From: nickd <ni...@gw...> - 2004-04-27 11:13:41
|
Hello list, I have been playing around with bacula for a little bit now. I am currently using AMANDA for backups and would like to switch to bacula, but there seems to be some performance issues that I am having. I seem to be able to backup using AMANDA quite a bit faster. I will be the first to admit user error, since I am new to this. I am just looking for some direction on what to check for. I have checked the archives and nothing seems to be improving my speeds. Here is some information on my setup. I have Bacula 1.34.2 running on FreeBSD 4.9-RELEASE-p3. Which also is the same machine that I have AMANDA on. They are both using the same attached Sony AIT 700C tape drive, I am however using differnt tapes with these tests. Nothing else of relevance is running on this machine when doing these tests. Average Tape Write Rates AMANDA 11148.9 KB/s Bacula 1189.7 KB/s Using btape's fill command I get rates like this. Begin writing Bacula records to first tape ... Block num = 0 Wrote block=5000, blk_num=5000 VolBytes=322,559,982 rate=16128.0 KB/s Wrote block=10000, blk_num=10000 VolBytes=645,119,926 rate=16128.0 KB/s Wrote block=15000, blk_num=15000 VolBytes=967,679,870 rate=16128.0 KB/s Flush block, write EOF Here is an snippet of my bacula-dir.conf. Director { Name = patsy-dir DIRport = 9101 QueryFile = "/home/bacula/conf/query.sql" WorkingDirectory = "/home/bacula/bin/working" PidDirectory = "/home/bacula/bin" Maximum Concurrent Jobs = 1 Password = "***" Messages = Standard } JobDefs { Name = "biddeford" Type = Backup Level = Incremental Client = patsy-fd FileSet = "Full Set" Schedule = "WeeklyCycle" Storage = SONY-700C Messages = Standard Pool = biddeford-pool Priority = 10 } Job { Name = "Client1" JobDefs = "biddeford" Write Bootstrap = "/home/bacula/bin/working/Client1.bsr" } Job { Name = "BackupCatalog" JobDefs = "biddeford" Level = Full FileSet="Catalog" Schedule = "WeeklyCycleAfterBackup" # This creates an ASCII copy of the catalog RunBeforeJob = "/home/bacula/conf/make_catalog_backup -u bacula" # This deletes the copy of the catalog RunAfterJob = "/home/bacula/conf/delete_catalog_backup" Write Bootstrap = "/home/bacula/bin/working/BackupCatalog.bsr" Priority = 11 } FileSet { Name = "Full Set" Include = signature=MD5 { /usr } Exclude = { /proc /tmp /.journal /.fsck } } Client { Name = patsy-fd Address = patsy.gwi FDPort = 9102 Catalog = MyCatalog Password = "***" File Retention = 30 days Job Retention = 6 months AutoPrune = yes } Storage { Name = SONY-700C Address = patsy.gwi SDPort = 9103 Password = "***" Device = SONY-700C Media Type = SONY-700C } Pool { Name = biddeford-pool Pool Type = Backup Recycle = yes AutoPrune = yes Volume Retention = 365 days Accept Any Volume = yes } (end snippet) I am currently using postgresql (also new to that) but was using sqlite with pretty much the same results in speed. Here is my latest output from a backup to tape. (Backup of /usr of the machine with the tape drive attached to it) patsy-dir: No prior Full backup Job record found. patsy-dir: No prior or suitable Full backup found. Doing FULL backup. patsy-dir: Start Backup JobId 9, Job=Client1.2004-04-26_13.02.07 patsy-sd: Wrote label to prelabeled Volume "monday0001" on device "/dev/sa0" patsy-dir: Bacula 1.34.2 (24Apr04): 26-Apr-2004 18:02 JobId: 9 Job: Client1.2004-04-26_13.02.07 Backup Level: Full (upgraded from Incremental) Client: patsy-fd FileSet: "Full Set" 2004-04-22 10:26:52 Start time: 26-Apr-2004 13:02 End time: 26-Apr-2004 18:02 FD Files Written: 198,773 SD Files Written: 198,773 FD Bytes Written: 1,778,758,033 SD Bytes Written: 1,805,112,561 Rate: 98.6 KB/s Software Compression: None Volume name(s): monday0001 Volume Session Id: 1 Volume Session Time: 1082998910 Last Volume Bytes: 1,812,563,582 Non-fatal FD errors: 0 SD Errors: 0 FD termination status: OK SD termination status: OK Termination: Backup OK patsy-dir: Begin pruning Jobs. patsy-dir: No Jobs found to prune. patsy-dir: Begin pruning Files. patsy-dir: No Files found to prune. patsy-dir: End auto prune. This backup took much longer then usual. Let me know what additional information anyone needs, any help or another direction to look would be greatly appreciated. Thanks in advance, Nick Duquette GWI Operations |