This chapter covers some common problems and error messages reported by NetWorker customers. Questions and answers were developed by Pat Darisme and Doug Murphy, with help from Sandeep Cariapa, Luc Chouinard, and Steve Modica.
The most common problem reported during NetWorker installations is that people forget that, in addition to installing NetWorker licenses with LicenseManager, you must also run nsrcap to initialize licenses. Without nsrcap, the following error message appears in the nwadmin window:
Server is disabled (Install base enabler) |
Run the nsrcap command to reread the license configuration file and enable the server.
# /usr/etc/nsrcap -v |
Q: The NetWorker server does not work, and the error message “No server available” keeps appearing. What can I do?
A: Check that all required subsystems are installed. See the IRIX NetWorker Installation Guide for a list of required server subsystems, and for a list of required client subsystems. After installing any new subsystems, stop and restart the NetWorker daemons, and reinitialize licensing information with nsrcap:
# /etc/init.d/networker stop # /etc/init.d/networker start # /usr/etc/nsrcap -v |
Q: How can I tell which licenses I have enabled on the machine?
A: In the nwadmin window, choose Registration from the Server menu. The Registration window appears, containing a list of enabled licenses.
Q: When I run nsrcap -v, the following error message appears:
nsrcap: RPC error, Program not registered |
A: You must launch the Networker daemons. The following command shows if the daemons are present:
# ps -ef | grep nsr root 704 527 0 Sep 14 ? 0:00 /usr/etc/nsrmmdbd root 527 1 0 Sep 14 ? 0:46 /usr/etc/nsrd root 543 1 0 Sep 14 ? 0:01 /usr/etc/nsrexecd |
If the NetWorker daemons are not present, you can start them with this command:
# /etc/init.d/networker start |
Finally, ensure that NetWorker daemons will be launched at boot time by doing:
# chkconfig networker on |
Q: When trying to license an autochanger (jukebox), this error message: appears:
The Jukebox "Jukebox Name" is not properly licensed Make sure the jukebox is properly enabled and authorized |
A: Using the LicenseManager command, install the appropriate autochanger license, and then rerun the nsrcap -v command.
Q: What is the difference between NetWorker server edition and network edition?
A: The server edition is intended for backing up a single machine, perhaps a database server. No client connections are allowed, meaning that server edition cannot back up other machines over the network. Server edition can be upgraded to network edition, which allows up to 10 client connections, meaning that NetWorker can back up 10 client machines over the network, or 9 plus the server itself.
Q: What is the difference between a client connection and ClientPak?
A: Client connections are licensed objects, whereas ClientPak is a software product distributed on CD-ROM. Out of the box, NetWorker network edition allows 10 client connections, allowing you to back up machines over the network, plus the server itself. Additional client connections can be purchased at any time, in units of 5, 25, and 100.
ClientPak products provide NetWorker clients for heterogeneous networks of machines, including most major hardware platforms and operating systems.
Q: Whenever I try to modify NetWorker settings I get this message:
Modify failed: user [email protected] needs to be on administrator list |
A: The root user (or other user on the administrator list for that NetWorker server) must add you to the list of privileged users. They must choose Server Setup from the Server menu, then modify the Administrator field.
Q: What licenses are available for NetWorker?
A: See the IRIX NetWorker Installation Guide for a list of NetWorker license options.
Q: I always see the following error messages when I launch NetWorker. Why?
Warning: Could not find Scheme file: SgiSpec Warning: Unmatched quotation marks in string ““, remaining fonts in list unparsed |
These messages are normal on systems without product image desktop_eoe.sw installed, and can be ignored.
Q: What device do I specify for backups?
A: The selection of an nrnsv (non-rewinding non-byte-swapping) device is critical. The nwadmin window's Devices panel often shows a nrnsv device selected, or if hardware compression is needed, a nrnsvc device. Thus the panel reads something like this:
/dev/rmt/tps1d6nrnsv 8mm 5GB (mounted) |
Use of a rewinding tape device can result in severe problems restoring backups, such as being able to restore only 3 MB of a 10 GB save set. If this accident occurs, your backup tapes are useless. Change the backup device by choosing Devices from the Media menu and clicking the Create button to designate an nrnsv device. Then relabel all tapes and perform a full backup as soon as possible.
Q: Although I know the tape is not full, when launching a backup I get this message:
No Space Left On Device |
A: Networker sometimes issues this message if it encounters write errors. This forces you to change tapes in order to ensure the safety of backups.
The probable cause of write errors is that the tape drive needs cleaning. A general rule of thumb is that a tape drive should be cleaned after every 30 to 50 hours of use.
Also bear in mind that the tape might actually be full. There is some overhead associated with Networker backups; in some cases it can be as much as 30%. This can be caused by empty tape space being used to keep the tape running at constant speed during backup, especially when the network is slow or for other reasons when data blocks arrive empty.
Q: Even though the tape is not write protected, when mounting a tape I get this message:
Tape write protected |
A: After failing to write data to a write-protected tape, NetWorker marks the tape in the media index as full. Even if the tape is nowhere near full, NetWorker says it is, as a way to deal with write protection. When you later make the tape physically read/write, you can no longer append to this tape, since the media index claims that it is full.
Fortunately the data on the tape is still readable and recoverable. Before proceeding, it would be wise to recover all data from this tape. Alternatively, you could retain the tape until all the save sets contained on it have expired.
The only way to use this tape again with Networker is to “recycle” it as a new tape. One way of doing this is to remove the index references to the tape so that NetWorker no longer thinks it is full. This can be done with the following command:
# nsrmm -dV volume |
This deletes the volume indexes, thus removing the volume from the media index and making the tape writable again. Warning: the command above makes it impossible to recover any existing save sets from that volume.
Q: Do tapes have to be relabeled before Networker starts rewriting to the new tapes?
A: Yes, relabeling a tape is an effective way of rewriting old tapes. However you can also recycle old tapes using the Volume Management window. Choose Volumes from the Media menu, then select a volume in the Volume Management window and choose Recycle from the Volume menu. This recycles a tape, allowing you to write to it again.
Q: In what format does NetWorker write data to the tape?
A: Networker writes data in XDR format (external data representation), with the media and file indexes before and after a save set. Thus tapes are readable only by NetWorker.
Q: What kind of autochanger can I use with Networker? Will my existing jukebox work with Networker?
A: Consult the NetWorker release notes for the currently supported list of jukeboxes, since this list is dynamic and depends on the version of NetWorker and IRIX that a system is running. Another way to obtain a list of supported jukeboxes is to run the jbm_enabler command, which outputs a list of jukeboxes that work, usually without any need to add special drivers into the kernel. The last entry is called “other” and may be used to configure custom support for hitherto unsupported jukebox.
If you have Web access, you can obtain the list of supported jukeboxes at this address: http://www.sgi.com/Products/hardware/challenge/Networker/NetWorker.html (look at the answers to frequent questions). Silicon Graphics recommends that only these supported devices be used with NetWorker. If you buy a jukebox from a third-party vendor, make sure that the firmware version matches the version recommended. Silicon Graphics supports jukeboxes from third-party vendors, if the firmware matches.
Q: When I run jbm_enabler, the following error message appears:
jbm_enabler: RPC error, Program not registered |
A: This usually means one of two things: either the NetWorker daemons are not running, or the customer has not installed (or has improperly installed) permanent licenses for the Autochanger option. Each jukebox requires a separate autochanger license. Autochanger licenses come in many sizes; see the marketing and product information for details. To check the NetWorker daemons, try the following test:
# ps -ef | grep nsr root 704 527 0 Sep 14 ? 0:00 /usr/etc/nsrmmdbd root 527 1 0 Sep 14 ? 0:46 /usr/etc/nsrd root 543 1 0 Sep 14 ? 0:01 /usr/etc/nsrexecd |
If the NetWorker daemons are not present, you can start them with this command:
# /etc/init.d/networker start |
If the daemons have already started, then check the /var/FLEXlm/license.dat file and verify that the licenses are installed and that they have not expired. Networker ships with temporary licenses for extra-cost options such as Jukebox and Archive, to assist you in evaluating the product.
A licensing problem typically results in this error message inside the nwadmin window:
The jukebox 'model_name' is not properly licensed. Make sure the jukebox is properly enabled and authorized. |
To help pinpoint a licensing problem, there are several things you can try. First run the jbexercise command to test jukebox connection and activity.
In the nwadmin window, choose Jukeboxes from the Media menu; then in the Jukeboxes window, choose Details from the View menu and verify that the Enabler code is present. Click the Label button and attempt jukebox labeling; helpful error messages may result. Try deleting and (re)creating the jukebox, by choosing Devices from the Media menu.
Once the jukebox is licensed, reinventory with the nsrjb -IE command.
Q: Where is the configuration file for my jukebox?
A: The jukebox configuration file /nsr/res/nsrjb.res contains licensing, scheduling, and current use information. This file is the roadmap for jukebox operation, and should not be modified, although sometimes it must be removed to force reinstallation of a jukebox (be sure to save a copy before removal).
Q: My jukebox is confused. It seems to expect some tapes to be mounted but cannot find them, or it continuously asks that I change tapes in the slots.
A: A jukebox's status can become confused for various reasons throughout operation. There are two ways around this. The first is to empty the jukebox and inform NetWorker that it is empty, then fill it up again. The second is to look at the nsrjb.res file, figure out what NetWorker expects in the jukebox, and put in what it wants.
The status of the jukebox is kept in the /nsr/res/nsrjb.res file, where two lines are critical to the state of the jukebox:
loaded slots: 5; loaded volumes: pipedream.005; |
These lines indicate that slot 5 is loaded and contains the volume pipedream.005. If this is not the actual state of the jukebox, then NetWorker gets confused. To alleviate this situation, stop the NetWorker daemons:
# /etc/init.d/networker stop |
Edit the two lines so that they read:
loaded slots: ; loaded volumes: ""; |
Remove tapes from the jukebox device, and restart the NetWorker daemons:
# /etc/init.d/networker start |
Finally, run nsrjb -IE to reinventory the jukebox and return it to a stable state.
To avoid having to do this repeatedly, you might want to make a copy of the nsrjb.res file, for example nsrjb.res.orig, when the jukebox is in a stable state. The next time the jukebox gets confused, replace the seemingly corrupted nsrjb.res file with the good copy, after stopping the NetWorker daemons:
# /etc/init.d/networker stop |
If you are familiar with Legato NetWorker on other platforms, note that the nsrjb -H option (hardware reset) is not supported in IRIX NetWorker. However, by opening the door of a jukebox or powering it down, you can get a similar effect.
Q: How do I change the cleaning slot on my jukebox?
A: Choose Jukeboxes from the Media menu. In the Jukeboxes window, change the slot number in the Cleaning slots field, and adjust the number of Available slots accordingly.
The cleaning slot entry corresponds to the loaded volumes entry in the nsrjb.res file. Here we see an error:
volumes: -, -, -, -, -, -, Cleaning Tape (5 uses left); ... cleaning slots: 1; |
In this case, either the 7th volume should be moved to first place, or the cleaning slot should say 7, depending on which hardware slot actually contains the cleaning tape.
Q: I started my backup last Friday. When I came in Monday, a message in the nwadmin window says NetWorker is waiting to mount a writable tape. What happened?
A: The advantage of a jukebox is that the administrator can set up a variety of backup schedules that do not require anybody to insert new tapes into a backup device. It is still possible to create schedules that require more tapes than a jukebox can hold. For answers on setting up schedules, see the section “Backup Schedules”.
Q: I had a jukebox license before, and I had to update it. Now when I run nsrcap -v, I get an error message saying the machine cannot find a valid license. What happened?
A: If you look at the /nsr/res/nsrjb.res file, you'll notice that it contains current license information for the jukebox. If the information in nsrjb.res does not match the license information in /var/flexlm/license.dat, then nsrjb.res wins. You might need to edit these to files by hand to bring them in sync. Then restart NetWorker and rerun nsrcap:
# /etc/init.d/networker start # /usr/etc/nsrcap -v |
Q: My jukebox cannot seem to mount volumes for backup or recovery. What's wrong?
A: If backup and recovery has worked in the past (you eliminated configuration errors), look at the /nsr/res/nsrjb.res file. Make sure entries in the file indicate all instructions the jukebox is getting from the NetWorker GUI tools or from the command line. Look for discrepancies between entries in this file and the perceived setup. This might give you some hints about what went wrong.
Q: I have 100 GB of data—why is a 7-slot DLT jukebox unable to hold this much data?
A: We talk more about compression directives later, but be aware that compression with NetWorker is optimized by choosing either software or hardware compression. If you try compressing files that have already been compressed, you will see a degradation in performance of your backup device. Also, make sure what you are backing up is in fact what you think you are backing up. Having directories NFS-mounted or CDROM drives on clients could possibly affect what NetWorker tries to back up.
Q: When should I inventory my jukebox? Should I ever reinventory my jukebox?
A: Choose Inventory from the Media menu whenever you have changed the set of tapes in the jukebox, or when NetWorker has trouble finding the correct volume to mount, which sometimes occurs if labels or indexes become corrupted. It never harms NetWorker to reinventory the jukebox.
Q: When performing a full backup, the jukebox loads tape 1, then tape 2, but then jumps to tape 7. What gives?
A: This is normal. Networker arbitrarily picks the order for writing tapes in a jukebox.
Q: What is the proper syntax for running the jbexercise command?
A: This command requires 2 arguments: the SCSI port of the jukebox controller, and the model number of the jukebox. To obtain a list of supported jukebox models, run the jbexercise command without arguments. Here is an example with proper syntax:
jbexercise -c /dev/scsi/sc6d6l0 -m "Silicon Graphics DLT2700" |
Q: When adding a jukebox, how do I know what control port the jukebox device is on? What is the “pathname of the control port” for a jukebox device?
A: Run the hinv command, which displays something like this:
Jukebox: unit 6, lun 1 on SCSI controller 5 |
This indicates that the jukebox device name would be /dev/scsi/sc5d6l0 (sc = SCSI controller, d = device unit, l = logical unit).
Q: How do I know the “pathname of media drive 1” to select a tape drive in the jukebox?
A: Looking at the /dev/rmt directory entries for the SCSI controller and device of the jukebox (with hinv output) you can pick the appropriate tape device. As per the previous example: tps5d6nrnsv or tps5d6nrnsvc for hardware compression.
Q: Why did NetWorker just disable my jukebox device?
A: Sometimes NetWorker disables a device after encountering an error. For instance, trying to move a tape into a full drive results in a disabled jukebox. To reset in nwadmin, choose Devices from the Media menu, and click the Enabled field's Yes button.
Q: In what format does NetWorker write data to the tape?
A: Networker writes data in XDR format (external data representation), with the media and file indexes before and after a save set. Thus tapes are readable only by NetWorker.
Q: Where can I get more information on DLT Jukeboxes?
A: Look at the Web site
http://www.sgi.com/Products/Mass_storage.html
Q: Where can I get more information on Exabyte Jukeboxes?
A: Look at the Web site http://www.sgi.com/Products/appsdirectory.dir/DeveloperIXExabyte_Corporation.html
Q: What's the difference between a level 5 and a level 9 backup?
A: A level 5 backs up more stuff, potentially. A level 1 backs up everything since the last full backup, even if level 1-9 backups have occurred in the meantime. A level 5 backs up everything to the last level 1-4 or full backup. Backup schedules using various levels are a good way to reduce the number of tapes needed for disaster recovery, while decreasing the frequency of time-consuming full backups. For more information about scheduling backups of different levels, see “Using Level Backups”.
Q: How many tapes do I need for a complete recovery of my client if it crashes?
A: That depends on the backup schedule you set for the client. The default schedule is for full backup on Sunday, and incrementals the remainder of the week. Under this schedule, the number of tapes needed for recovery increases each day, from 1 on Sunday to 7 on Saturday. (Incrementals could be contained on the same tape, although this provides poor protection against fire). If you set a schedule with full backup on Sunday, and level 9 backups the remainder of the week, the number of tapes needed for recovery should never exceed 2. (Of course level 9 backups would take longer than incrementals, especially at the end of the week.)
Q: How long should it take for NetWorker to backup my clients?
A: This depends on the device you use, how your schedules are staggered, the network speed, parallelism, and other factors. If you assume a rate of 1MB/second for a DLT drive, then it should take close to 3 hours to backup 10GB of data. If you find that your backups are taking too long, then investigate factors such as network speed, CPU speed, tape drive speed, NetWorker parallelism, and concurrent devices.
Q: What is the difference between Indexes and Save Sets?
A: An index is a browsable directory tree indicating which files are available for recovery. Although they are copied to tape, browsable indexes are stored on server disk, and thus do not need to be retrieved from tape. They contain information about files backed up for each NetWorker client. A save set is a bunch of files backed up on tape.
Since indexes consume lots of disk space, they are typically browsable (stored on disk) for much less time than save sets are retained on tape. The browse policy governs the longevity of client indexes, whereas the retention policy controls the minimum age of save sets when a tape gets marked as recyclable.
Q: How do I define browse and save set retention policies?
A: Both the browse policy and the retention policy may be set by choosing Client Setup from the Clients menu. The default browse period is one month; the default retention period is one year. Although there is seldom a need for doing so, policy periods may be set by choosing Policies from the Customize menu.
Q: How can I keep a backup longer than the retention policy? What is archiving?
A: You can make a clone of a backup volume by choosing Clone Volume from the Media menu. You can make a clone of a save set by choosing Clone from the Save Set menu. Cloned volumes and save sets can be stored forever, and recovered by means of the scanner command.
Archiving involves the long-term storage of data on media called archive volumes. Archive volumes are like backup volumes, except their retention period is unlimited, and archives are not tracked in the file index. Archiving requires an Archive license, and works with either server or network edition.
Q: Where do I define which directories to back up? How can I control the contents of a save set?
A: Each client's save sets may be specified by choosing Client Setup from the Clients menu. In the Save set field, indicate which directories and/or files you wish to have backed up on that client.
Q: What's the difference between the media index and the file index?
A: The media index contains information about tapes, whereas the file index contains information about backed-up files and directories. As long as files remain in the online file index, they are browsable and recoverable from the nwrecover window.
The media index works in tandem with the file index to tell NetWorker which volume contains a particular saveset. As long as volume entries remain in the media index, save sets on that volume are recoverable by choosing Recover from the nwadmin Save Set menu. Both the media index and the file index exist online, and are also copied to tape (twice) during each scheduled backup.
Q: What does labeling a tape do?
A: The tape label is how Networker (and the operator) identify a tape and determine which save sets reside on which tape. The physical label and the magnetic label should match. When you relabel a tape, you remove all entries for that volume from the media index. This makes it impossible to recover data from that tape using the Save Set Recover window.
When a tape is marked recyclable, you must relabel it before overwriting it.
Q: How can I check the status of a magnetic tape label? How do I verify a label?
A: To verify a tape label, or to see if it is corrupted, try the following command:
# nsrmm -v -p -f /dev/rmt/tpsXdYnrnsv |
Replace X with the SCSI controller number, and Y with the device number of the drive. If the label is good, nsrmm should display output similar to the following:
/dev/rmt/tps0d3nrnsv: verifying label, moving backward 2 files /dev/rmt/tps0d3nrnsv: verified label of vol_test2 Verified 4mm tape vol_test2 on /dev/rmt/tps0d3nrnsv expiration time is Tue Jun 16 16:43:52 1998 |
Q: I am unable to launch a group backup from the nwadmin window. Why?
A: Try launching this backup from the command line by typing:
# savegroup group_name
|
This might provide debugging information that can help you isolate the problem.
Q: How do I force an immediate launch of a backup without waiting for the scheduled time to arrive?
A: In the nwadmin window, click the Group Control button (or choose Group Control from the Server menu), select a group from the Groups list, and click the Start button.
Q: I keep getting the following error message. Why?
Permission denied... backup failed. |
A: The client needs to be informed that the NetWorker server needs to read its files. On the client system, run chkconfig networker on, which starts the nsrexecd daemon to perform backups. It is possible to modify to modify /.rhosts to grant NetWorker server access, but this method is less secure than nsrexecd.
Q: I keep getting the following error message. Why?
root is not an authorized user. |
In the Remote access field, you also need to add the username of the agent (probably root) who runs the backup remotely.
Q: When trying to recover files, I get the following error message. Why?
hostname is not a registered client. |
A: Automatic server detection has located a server on which the current system is not a registered client. If a different server is appropriate, run the command again with the -s option to specify server. If the server is correct, register that client on the server. In nwadmin, choose Client Setup from the Clients menu. In the Clients window, click Create, and type the client name in the Name field.
Q: I keep getting the following error message. Why?
user root on machine server is not on server.FQDN's remote access list |
This is usually the result of not having both a bare hostname and a fully qualified domain name (FQDN) for the NetWorker server. In nwadmin, choose Client Setup from the Clients menu, select the server in the Clients window, scroll down to the Aliases field, and add the bare hostname as an alias for the server. If the aliasing information is correct, check the Remote access field for accuracy.
Q: Where do I find the browsable indexes?
A: If indexes are still browsable, you can see them by choosing Indexes from the Clients menu in the nwadmin window, and users can browse them in the nwrecover window.
Q: How do I recover files if the file index gets erased or their browse period expires?
A: The media index, which is not affected by the browse period, contains information about which save sets are on each volume. You can browse the media index by choosing Volumes from the Media menu; in the Volume Management window, locate the save set you want. Then choose Recover from the Save Set menu; in the Save Set Recover window, select the correct version and click the Recover button.
Q: How can I recover files without using nwrecover?
A: To avoid the nwrecover window, you can use the mminfo and recover commands in combination. First run mminfo -v to print the Save Set IDs (ssid) list, and choose the one you wish to recover. Then run recover -S ssid to interactively recover files.
# mminfo -v volume client date time size ssid fl lvl name test.006 envlab48 9/23/96 20:16:58 31 MB 12821 cb full /n/nw-test-dir # recover -s servername -S 12821 |
You can recover specific directories from specific savesets without going through the recover interface by using the scanner command:
# scanner -s ssid /dev/rmt/tpsXdXnrnsv -x uasm -vr /full/path/dirname |
You can rename the recovery directory by specifying -m /full=/fullnew (for example). You can also use scanner to recover individual files, or the Save Set Recover feature of nwadmin. Before recovering any client files, run nsrck first. If there are no problems with the index, it should exit quickly:
# nsrck clientname |
Q: The mminfo -v command displays nine fields: volume, client, date, time, size, ssid, fl, lvl, and name. What does the “fl” field indicate, and what do its values represent?
A: The “fl” (flags) field indicates whether a save set spans multiple volumes.
cr | complete, recoverable |
hr | just the head, recoverable |
tr | just the tail, recoverable |
Q: How can I recover a single file using nwadmin (instead of nwrecover)?
A: In nwadmin, choose Recover from the Save Set menu.
In the Save Set Recover window, click the Recover button (at the bottom right). In the Save Set Recover Status window, type the full path name of the file that you wish to recover in the Paths to recover field, then press [Enter] to register your selection. If you wish to relocate or overwrite the file, click the Options button. Finally, click the Start button to initiate recovery.
Q: When a file is recovered, does its modification time get changed?
A: No, a recovered file retains the modification time it had when it was backed up.
Q: How can I determine which save sets are on a tape?
A: To list the save sets on a tape, run the scanner command:
# scanner /dev/rmt/tpsXdXnrnsv |
Q: How could I list the files in a particular save set, such as #8738783?
A: Run the scanner -s command, piped to the uasm command with -rvn option:
# scanner -s 8738783 /dev/rmt/tpsXdXnrnsv | uasm -rvn |
Q: How do I rebuild the online file index for a save set, such as #8738783?
A: Run the scanner command with -i and -s options:
# scanner -i -s 8738783 /dev/rmt/tpsXdXnrnsv |
Q: How can I recover files if the media index has been corrupted or removed?
A: First get the bootstrap ID of the most recent full backup (use the mminfo -v command) then rebuild the media index (use the scanner -i command, which rebuilds both online file and media indexes). Of course recent changes to files will be lost.
Q: Can I recover files after their retention period has expired?
A: Perhaps, depending on how you set up schedules. Normally NetWorker does not reuse volumes until one backup cycle has completed. So before giving up, check your most recent backups volumes and you might find some tape labels that have not yet been rewritten. Of course, once a tape has been relabeled, not just marked for recycling, old data is no longer recoverable. Browse files in the save set to determine if you wish to recover them. Use the recover -a command to force an automatic (non-interactive) recover of the entire save set without prompting for each file.
Q: Where is the documentation for using directives?
A: The nsr(4) reference page contains complete information about directives.
Q: What is a directive?
A: A directive is an improved method for backing up special files. Directives may be stored in the .nsr file of any directory, and are parsed by NetWorker commands, called ASMs (application specific modules), to determine how to save that directory. If there are no directives, everything in the directory gets saved.
Q: What is the correct syntax for a .nsr directive file?
A: There are three forms, as described on the nsr(4) reference page. The first is a particular ASM specification. The second is a “save environment” directive. The third is a <<dir>> directive. A # at the beginning of a line indicates a comment.
[+] ASM [args ...] : pattern ... save environment << dir >> |
In the first line the colon (:) is an internal field separator between the ASM specification and its arguments. As usual, patterns requiring special characters must be enclosed in double quotes ("). The following patterns are illegal: ".." and "/". This is because all patterns must match the current directory. The optional [+] key means the directive file is propagated to subdirectories. The save environment and <<dir>> directives are both adequately explained by the nsr(4) reference page.
Q: What ASMs (application specific modules) are available with NetWorker?
A: The principal NetWorker ASM is uasm, the module for saving and recovering UNIX filesystem data. You can use uasm manually during save and recover, for instance where you extract a particular file from piped output. For example, this command extracts a save set for client mars and relocates it from root to the /a filesystem:
# scanner -c mars /dev/rmt/tps3d4nrnsv | uasrm -rv -m /=/a |
See the uasm(1M) reference page for more information about uasm, and for descriptions of the following ASMs: skip, null, nullasm, holey, always, logasm, mailasm, atimeasm, tarasm, swapasm, xlateasm, compressasm, nsrmmdbasm, nsrindexasm.
All these have the same basic options, so please refer to the reference pages when determining which ones to use for complicated configurations.
Q: Where is the compression being done on my backups?
A: There are only two ways of compressing backups: in the software using compression directives (compressasm) defined in a .nsr file, or in the hardware using the tape drive's firmware implementation of compression.
Q: How do I enable hardware (H/W) compression?
A: For hardware compression you must specifically use the “c” (compression) device. For instance, specify /dev/rmt/tpsXdYnrnsvc instead of /dev/rmt/tpsXdYnrnsv.
There are some NetWorker configuration changes to support hardware compression. You must inform Networker that tape data capacity has been enlarged. In the nwadmin window, choose Devices from the Media menu. Then in the Devices window, choose Details from the View menu.
For a DLT device, select the hardware compression drive in the Devices field, and enter “20GB” in the Volume default capacity field. Note that NetWorker does not use this value until it labels a tape, so it does not take effect immediately.
Some words about compression: software compression is done on the client during backup, whereas hardware compression is done on the server. This has implications for network throughput, drive throughput, CPU usage, and so forth.
Q: What is the compression rate?
A: The Legato compressasm uses their implementation of Lempel-Ziv coding to compress file pages (one page is always 8 KB) while saving them. This can result in less data being transmitted over the network and less data being stored. Since network and media subsystems are normally major performance bottlenecks, compression provides significant performance improvement, especially if the client processor is relatively fast. Compression of 50-60% is typical, but results vary with file content. Already-compressed files, such as GIF or JPEG images, do not benefit further from either software or hardware compression.
Q: How can I see a log of NetWorker activities?
A: People often remark that information in the Messages panel of the nwadmin window does not go back far enough. For NetWorker server messages going back to when you started the networker daemons, see the standard log file /nsr/logs/daemon.log.
If you want to save daemon log files from previous NetWorker starts, you can change the /etc/init.d/networker file to save the old log file before starting nsrd.
Q: I no longer receive mail notification for savegroup completion. Why?
A: Verify that mail for root is working.Mail failure is the likely cause of this problem.