Re: [bareos-users] losing LTO3 tape capacity with bareos. But full tape length can be written with dd or btape speed
Hi Tim, if dd use the full speed instead of bareos, whats about an IO-limitation on the source - the spool-disk? In your example, the first 50-gb spoolfile was transferred with 40MB/s the later ones with 35MB/s. Does the spool-disks was filled with other backup-jobs (writes) during the despooling? What kind of spool-disks are in use (raid level, how much disks, what kind of disks/ssd)? I had an high load on the backup-server with an spooldisk (for 3 lto-4 drives) with raid10 on 6*SAS 15k disks. With raid0 (the spool disks only) the load was ok. Udo On 21.08.2018 15:46, Tim Banchi wrote: > Hello, > > I'm using Bareos 17.2 with the following pool, and device configuration: > > Pool { > Name = tape_automated > Pool Type = Backup > Recycle = yes # Bareos can automatically recycle > Volumes > AutoPrune = yes # Prune expired volumes > #Recycle Oldest Volume = yes > RecyclePool = Scratch > Maximum Volume Bytes = 0 > Job Retention = 365 days > Volume Retention = 4 weeks > Volume Use Duration = 12 days > Cleaning Prefix = "CLN" > Catalog Files = yes > Storage = delaunay_HP_G2_Autochanger #virtualfull test > # Next Pool = tape_automated > } > > Device { > Name = "Ultrium920" > Media Type = LTO > Archive Device = /dev/st1 > Autochanger = yes > LabelMedia = no > AutomaticMount = yes > AlwaysOpen = yes > RemovableMedia = yes > Maximum Spool Size = 50G > Spool Directory = /var/lib/bareos/spool > Maximum Block Size = 2097152 > # Maximum Block Size = 4194304 > Maximum Network Buffer Size = 32768 > Maximum File Size = 50G > Alert Command = "/bin/bash -c '/usr/sbin/smartctl -H -l error %c'" > } > > I'm closely monitoring tape performance (write speed + capacity marked as > full by bareos), and all tapes are losing capacity, and write speed, in an > almost linear way. > > Over the period of roughly 8 months (I installed bareos early this year) and > 3 to 4 write cycles per tape (=every 2 months a tape is rewritten), I've lost > on average 25% of the initial capacity per tape. E.g. some tapes started at > ~400GB, and are now somewhere at 300GB. some started at 160GB, and are now > somewhere around 120GB. > > Write-speed wise, I also lose speed: The fastest tapes (most often also the > tapes which have the highest capacity) started out at 50MB/s, and are now > somewhere at 42MB/s. The worst tapes are somewhere between 20MB/s, and are > now at 16MB/s. > > Because most of the tapes were second-hand (we got tape drive, controller, > autoloader, and tapes as a gift, as our NGO doesn't have any money), I blamed > the bad state of tapes (I verify volume to catalog after each job, I assume > I'm still fine. Also there are no tape alerts). > But in the course of this year, I also introduced some brand new (never > written) tapes. And they are also losing capacity, and write speed to the > same amount. > > Hardware compression, (software) encryption, and software compression is all > ON since the very beginning (so no configuration change). And I don't think > this can be relevant, because tapes are losing capacity over time, whether > they started off at ~406GB (new) or 160GB (used, and initial bad condition). > It also doesn't depend what I back up (everything is encrypted anyway, so > compression shouldn't work at all at the drive's side). > > I don't know the logic how bareos recognises that a tape is full, so I > thought I give it a try to fill some tapes with zero/urandom to "realign" it. > I could fully write to those tapes (=400GB). But bareos continued to only > partly use them as before. I then tried to fill the tapes up with btape > speed, but the same problem persists when using the tape again in bareos. > > Tape backups typically have job logs like: > > *Joblog* > Connecting to Director localhost:9101 > 1000 OK: pavlov-dir Version: 17.2.4 (21 Sep 2017) > Enter a period to cancel a command. > list joblog jobid=7665 > Automatically selected Catalog: MyCatalog > Using Catalog "MyCatalog" > 2018-08-18 06:37:51 pavlov-dir JobId 7665: Start Backup JobId 7665, > Job=edite_backup_automated.2018-08-18_05.30.00_34 > 2018-08-18 06:37:52 pavlov-dir JobId 7665: Using Device "Ultrium920" to > write. > 2018-08-18 06:37:53 delaunay-sd JobId 7665: 3307 Issuing autochanger "unload > slot 2, drive 0" command. > 2018-08-18 06:40:22 delaunay-sd JobId 7665: 3304 Issuing autochanger "load > slot 1, drive 0" command. > 2018-08-18 06:41:06 delaunay-sd JobId 7665: 3305 Autochanger "load slot 1, > drive 0", status is OK. > 2018-08-18 06:41:15 delaunay-sd JobId 7665: Volume "XM2705L3" previously > written, moving to end of data. > 2018-08-18 06:41:33 delaunay-sd JobId 7665: Ready to append to end of Volume > "XM2705L3" at file=5. > 2018-08-18 06:41:33 delaunay-sd JobId 7665: Spooling data ... > 2018-08-18 08:37:57 delaunay-sd JobId 7665: User specified Device spool size > reached: DevSpoolSize=53,687,398,
Re: [bareos-users] Bareos Wild and Regex
No luck. If I put a separate include with it's options Bareos still will not exclude the directory with a WildDir or RegExDir. Given that bregex and bwild work just fine I think there is something broken somewhere in my config but I have no idea where. Bareos just won't exclude that directory. -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [bareos-users] Bareos Wild and Regex
Interesting. I may split it into two includes with one having exclude = yes in it. I think it's Bareos acting wonky since bregex and bwild do the selections perfectly no matter what depth the directory is. So something must be happening when it gets parsed. I notice you put your Files = after the includes. Any reason since I assume the config engine doesn't care. -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to bareos-users+unsubscr...@googlegroups.com. To post to this group, send email to bareos-users@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [bareos-users] losing LTO3 tape capacity with bareos. But full tape length can be written with dd or btape speed
Hi Kai, I thought about disabling hardware compression, but doing that in a permanent way, I think I would have had to re-install a OS which runs LTT (other ways via Linux seemed to me not conclusive enough, or I didn't find that information). I also read at numerous occasions that >= LTO-2 drives handle hardware compression quite well (e.g. not increasing data usage when compressing uncompressable data like encrypted data). ad software compression: we (have to) use quite a slow internal network, so software compression at the client seemed reasonable to me. On Wednesday, August 22, 2018 at 12:22:20 PM UTC+2, Stefan Klatt wrote: > Hi Tim, > > > > hmmm. "Hardware compression, (software) encryption, and software > compression" > > That's probably overkill. Could you disable hardware compression? > This can slow down backup if the data is encrypted (and compressed). > > > > Stefan > > > > > Am 22.08.2018 um 09:40 schrieb Tim > Banchi: > > > > Hi Stefan, > > yes, I use newly purchased ones when the autoloader/drive requested it. I got > clean requests when getting I/O errors with really bad tapes (which I then > dumped right away). Otherwise there have been so far no cleaning requests > (which is normal to my experience), and the status of the drive is OK. I also > looked up regularly logged warnings/errors, and there is nothing (except the > aforementioned I/O errors) > > > > On Tuesday, August 21, 2018 at 10:38:26 PM UTC+2, Stefan Klatt wrote: > > > Hi Tim, > > > > only a idea... did you use cleaning tapes? > > > > Regards > > > > Stefan > > > > > Am 21.08.2018 um 15:46 schrieb Tim > Banchi: > > > > Hello, > > I'm using Bareos 17.2 with the following pool, and device configuration: > > Pool { > Name = tape_automated > Pool Type = Backup > Recycle = yes # Bareos can automatically recycle > Volumes > AutoPrune = yes # Prune expired volumes > #Recycle Oldest Volume = yes > RecyclePool = Scratch > Maximum Volume Bytes = 0 > Job Retention = 365 days > Volume Retention = 4 weeks > Volume Use Duration = 12 days > Cleaning Prefix = "CLN" > Catalog Files = yes > Storage = delaunay_HP_G2_Autochanger #virtualfull test > # Next Pool = tape_automated > } > > Device { > Name = "Ultrium920" > Media Type = LTO > Archive Device = /dev/st1 > Autochanger = yes > LabelMedia = no > AutomaticMount = yes > AlwaysOpen = yes > RemovableMedia = yes > Maximum Spool Size = 50G > Spool Directory = /var/lib/bareos/spool > Maximum Block Size = 2097152 > # Maximum Block Size = 4194304 > Maximum Network Buffer Size = 32768 > Maximum File Size = 50G > Alert Command = "/bin/bash -c '/usr/sbin/smartctl -H -l error %c'" > } > > I'm closely monitoring tape performance (write speed + capacity marked as > full by bareos), and all tapes are losing capacity, and write speed, in an > almost linear way. > > Over the period of roughly 8 months (I installed bareos early this year) and > 3 to 4 write cycles per tape (=every 2 months a tape is rewritten), I've lost > on average 25% of the initial capacity per tape. E.g. some tapes started at > ~400GB, and are now somewhere at 300GB. some started at 160GB, and are now > somewhere around 120GB. > > Write-speed wise, I also lose speed: The fastest tapes (most often also the > tapes which have the highest capacity) started out at 50MB/s, and are now > somewhere at 42MB/s. The worst tapes are somewhere between 20MB/s, and are > now at 16MB/s. > > Because most of the tapes were second-hand (we got tape drive, controller, > autoloader, and tapes as a gift, as our NGO doesn't have any money), I blamed > the bad state of tapes (I verify volume to catalog after each job, I assume > I'm still fine. Also there are no tape alerts). > But in the course of this year, I also introduced some brand new (never > written) tapes. And they are also losing capacity, and write speed to the > same amount. > > Hardware compression, (software) encryption, and software compression is all > ON since the very beginning (so no configuration change). And I don't think > this can be relevant, because tapes are losing capacity over time, whether > they started off at ~406GB (new) or 160GB (used, and initial bad condition). > It also doesn't depend what I back up (everything is encrypted anyway, so > compression shouldn't work at all at the drive's side). > > I don't know the logic how bareos recognises that a tape is full, so I > thought I give it a try to fill some tapes with zero/urandom to "realign" it. > I could fully write to those tapes (=400GB). But bareos continued to only > partly use them as before. I then tried to fill the tapes up with btape > speed, but the same problem persists when using the tape again in bareos. > >
Re: [bareos-users] losing LTO3 tape capacity with bareos. But full tape length can be written with dd or btape speed
Hi Kai, thanks, the HP LTT tools contain a diagnostic test (my drive is a HP Ultrium920 SCSI LTO3 drive). There should also be a 5-second self test on every power up. So far, this always succeeded (I never got a error message on the autoloader display or in the logs). I remember, that I did this LTT diagnostic when configuring the drive in January. However, right now, I don't have a access to a physical Windows Server any more where I could try that quickly out. On Wednesday, August 22, 2018 at 10:12:23 AM UTC+2, Kai Zimmer wrote: > Hi Tim, > > does the drive have a self test function in the firmware? If so, try it. > If not, maybe there is a test software from the vendor available? > > Best regards, > Kai > > Am 22.08.2018 um 09:40 schrieb Tim Banchi: > > Hi Stefan, > > > > yes, I use newly purchased ones when the autoloader/drive requested it. I > > got clean requests when getting I/O errors with really bad tapes (which I > > then dumped right away). Otherwise there have been so far no cleaning > > requests (which is normal to my experience), and the status of the drive is > > OK. I also looked up regularly logged warnings/errors, and there is nothing > > (except the aforementioned I/O errors) > > > > > > > > On Tuesday, August 21, 2018 at 10:38:26 PM UTC+2, Stefan Klatt wrote: > >> Hi Tim, > >> > >> > >> > >> only a idea... did you use cleaning tapes? > >> > >> > >> > >> Regards > >> > >> > >> > >> Stefan > >> > >> > >> > >> > >> Am 21.08.2018 um 15:46 schrieb Tim > >>Banchi: > >> > >> > >> > >>Hello, > >> > >> I'm using Bareos 17.2 with the following pool, and device configuration: > >> > >> Pool { > >>Name = tape_automated > >>Pool Type = Backup > >>Recycle = yes # Bareos can automatically recycle > >> Volumes > >>AutoPrune = yes # Prune expired volumes > >>#Recycle Oldest Volume = yes > >>RecyclePool = Scratch > >>Maximum Volume Bytes = 0 > >>Job Retention = 365 days > >>Volume Retention = 4 weeks > >>Volume Use Duration = 12 days > >>Cleaning Prefix = "CLN" > >>Catalog Files = yes > >>Storage = delaunay_HP_G2_Autochanger #virtualfull test > >> # Next Pool = tape_automated > >> } > >> > >> Device { > >>Name = "Ultrium920" > >>Media Type = LTO > >>Archive Device = /dev/st1 > >>Autochanger = yes > >>LabelMedia = no > >>AutomaticMount = yes > >>AlwaysOpen = yes > >>RemovableMedia = yes > >>Maximum Spool Size = 50G > >>Spool Directory = /var/lib/bareos/spool > >>Maximum Block Size = 2097152 > >> # Maximum Block Size = 4194304 > >>Maximum Network Buffer Size = 32768 > >>Maximum File Size = 50G > >>Alert Command = "/bin/bash -c '/usr/sbin/smartctl -H -l error %c'" > >> } > >> > >> I'm closely monitoring tape performance (write speed + capacity marked as > >> full by bareos), and all tapes are losing capacity, and write speed, in an > >> almost linear way. > >> > >> Over the period of roughly 8 months (I installed bareos early this year) > >> and 3 to 4 write cycles per tape (=every 2 months a tape is rewritten), > >> I've lost on average 25% of the initial capacity per tape. E.g. some tapes > >> started at ~400GB, and are now somewhere at 300GB. some started at 160GB, > >> and are now somewhere around 120GB. > >> > >> Write-speed wise, I also lose speed: The fastest tapes (most often also > >> the tapes which have the highest capacity) started out at 50MB/s, and are > >> now somewhere at 42MB/s. The worst tapes are somewhere between 20MB/s, and > >> are now at 16MB/s. > >> > >> Because most of the tapes were second-hand (we got tape drive, controller, > >> autoloader, and tapes as a gift, as our NGO doesn't have any money), I > >> blamed the bad state of tapes (I verify volume to catalog after each job, > >> I assume I'm still fine. Also there are no tape alerts). > >> But in the course of this year, I also introduced some brand new (never > >> written) tapes. And they are also losing capacity, and write speed to the > >> same amount. > >> > >> Hardware compression, (software) encryption, and software compression is > >> all ON since the very beginning (so no configuration change). And I don't > >> think this can be relevant, because tapes are losing capacity over time, > >> whether they started off at ~406GB (new) or 160GB (used, and initial bad > >> condition). It also doesn't depend what I back up (everything is encrypted > >> anyway, so compression shouldn't work at all at the drive's side). > >> > >> I don't know the logic how bareos recognises that a tape is full, so I > >> thought I give it a try to fill some tapes with zero/urandom to "realign" > >> it. I could fully write to those tapes (=400GB). But bareos continued to > >> only partly use them as before. I then tried to fill the tapes up with > >> btape speed, but the same problem p
Re: [bareos-users] losing LTO3 tape capacity with bareos. But full tape length can be written with dd or btape speed
Hi Tim, hmmm. "Hardware compression, (software) encryption, and software compression" That's probably overkill. Could you disable hardware compression? This can slow down backup if the data is encrypted (and compressed). Stefan Am 22.08.2018 um 09:40 schrieb Tim Banchi: > Hi Stefan, > > yes, I use newly purchased ones when the autoloader/drive requested it. I got > clean requests when getting I/O errors with really bad tapes (which I then > dumped right away). Otherwise there have been so far no cleaning requests > (which is normal to my experience), and the status of the drive is OK. I also > looked up regularly logged warnings/errors, and there is nothing (except the > aforementioned I/O errors) > > > > On Tuesday, August 21, 2018 at 10:38:26 PM UTC+2, Stefan Klatt wrote: >> Hi Tim, >> >> >> >> only a idea... did you use cleaning tapes? >> >> >> >> Regards >> >> >> >> Stefan >> >> >> >> >> Am 21.08.2018 um 15:46 schrieb Tim >> Banchi: >> >> >> >> Hello, >> >> I'm using Bareos 17.2 with the following pool, and device configuration: >> >> Pool { >> Name = tape_automated >> Pool Type = Backup >> Recycle = yes # Bareos can automatically recycle >> Volumes >> AutoPrune = yes # Prune expired volumes >> #Recycle Oldest Volume = yes >> RecyclePool = Scratch >> Maximum Volume Bytes = 0 >> Job Retention = 365 days >> Volume Retention = 4 weeks >> Volume Use Duration = 12 days >> Cleaning Prefix = "CLN" >> Catalog Files = yes >> Storage = delaunay_HP_G2_Autochanger #virtualfull test >> # Next Pool = tape_automated >> } >> >> Device { >> Name = "Ultrium920" >> Media Type = LTO >> Archive Device = /dev/st1 >> Autochanger = yes >> LabelMedia = no >> AutomaticMount = yes >> AlwaysOpen = yes >> RemovableMedia = yes >> Maximum Spool Size = 50G >> Spool Directory = /var/lib/bareos/spool >> Maximum Block Size = 2097152 >> # Maximum Block Size = 4194304 >> Maximum Network Buffer Size = 32768 >> Maximum File Size = 50G >> Alert Command = "/bin/bash -c '/usr/sbin/smartctl -H -l error %c'" >> } >> >> I'm closely monitoring tape performance (write speed + capacity marked as >> full by bareos), and all tapes are losing capacity, and write speed, in an >> almost linear way. >> >> Over the period of roughly 8 months (I installed bareos early this year) and >> 3 to 4 write cycles per tape (=every 2 months a tape is rewritten), I've >> lost on average 25% of the initial capacity per tape. E.g. some tapes >> started at ~400GB, and are now somewhere at 300GB. some started at 160GB, >> and are now somewhere around 120GB. >> >> Write-speed wise, I also lose speed: The fastest tapes (most often also the >> tapes which have the highest capacity) started out at 50MB/s, and are now >> somewhere at 42MB/s. The worst tapes are somewhere between 20MB/s, and are >> now at 16MB/s. >> >> Because most of the tapes were second-hand (we got tape drive, controller, >> autoloader, and tapes as a gift, as our NGO doesn't have any money), I >> blamed the bad state of tapes (I verify volume to catalog after each job, I >> assume I'm still fine. Also there are no tape alerts). >> But in the course of this year, I also introduced some brand new (never >> written) tapes. And they are also losing capacity, and write speed to the >> same amount. >> >> Hardware compression, (software) encryption, and software compression is all >> ON since the very beginning (so no configuration change). And I don't think >> this can be relevant, because tapes are losing capacity over time, whether >> they started off at ~406GB (new) or 160GB (used, and initial bad condition). >> It also doesn't depend what I back up (everything is encrypted anyway, so >> compression shouldn't work at all at the drive's side). >> >> I don't know the logic how bareos recognises that a tape is full, so I >> thought I give it a try to fill some tapes with zero/urandom to "realign" >> it. I could fully write to those tapes (=400GB). But bareos continued to >> only partly use them as before. I then tried to fill the tapes up with btape >> speed, but the same problem persists when using the tape again in bareos. >> >> Tape backups typically have job logs like: >> >> *Joblog* >> Connecting to Director localhost:9101 >> 1000 OK: pavlov-dir Version: 17.2.4 (21 Sep 2017) >> Enter a period to cancel a command. >> list joblog jobid=7665 >> Automatically selected Catalog: MyCatalog >> Using Catalog "MyCatalog" >> 2018-08-18 06:37:51 pavlov-dir JobId 7665: Start Backup JobId 7665, >> Job=edite_backup_automated.2018-08-18_05.30.00_34 >> 2018-08-18 06:37:52 pavlov-dir JobId 7665: Using Device "Ultrium920" to >> write. >> 2018-08-18 06:37:53 delaunay-sd JobId 7665: 3307 Issuing autochanger >> "unload slot 2, drive 0" command. >> 2018-08-18 06:40:22 delaunay-sd JobId 7665: 3304 Issuing autochanger "load >> sl
Re: [bareos-users] losing LTO3 tape capacity with bareos. But full tape length can be written with dd or btape speed
Hi Tim, does the drive have a self test function in the firmware? If so, try it. If not, maybe there is a test software from the vendor available? Best regards, Kai Am 22.08.2018 um 09:40 schrieb Tim Banchi: Hi Stefan, yes, I use newly purchased ones when the autoloader/drive requested it. I got clean requests when getting I/O errors with really bad tapes (which I then dumped right away). Otherwise there have been so far no cleaning requests (which is normal to my experience), and the status of the drive is OK. I also looked up regularly logged warnings/errors, and there is nothing (except the aforementioned I/O errors) On Tuesday, August 21, 2018 at 10:38:26 PM UTC+2, Stefan Klatt wrote: Hi Tim, only a idea... did you use cleaning tapes? Regards Stefan Am 21.08.2018 um 15:46 schrieb Tim Banchi: Hello, I'm using Bareos 17.2 with the following pool, and device configuration: Pool { Name = tape_automated Pool Type = Backup Recycle = yes # Bareos can automatically recycle Volumes AutoPrune = yes # Prune expired volumes #Recycle Oldest Volume = yes RecyclePool = Scratch Maximum Volume Bytes = 0 Job Retention = 365 days Volume Retention = 4 weeks Volume Use Duration = 12 days Cleaning Prefix = "CLN" Catalog Files = yes Storage = delaunay_HP_G2_Autochanger #virtualfull test # Next Pool = tape_automated } Device { Name = "Ultrium920" Media Type = LTO Archive Device = /dev/st1 Autochanger = yes LabelMedia = no AutomaticMount = yes AlwaysOpen = yes RemovableMedia = yes Maximum Spool Size = 50G Spool Directory = /var/lib/bareos/spool Maximum Block Size = 2097152 # Maximum Block Size = 4194304 Maximum Network Buffer Size = 32768 Maximum File Size = 50G Alert Command = "/bin/bash -c '/usr/sbin/smartctl -H -l error %c'" } I'm closely monitoring tape performance (write speed + capacity marked as full by bareos), and all tapes are losing capacity, and write speed, in an almost linear way. Over the period of roughly 8 months (I installed bareos early this year) and 3 to 4 write cycles per tape (=every 2 months a tape is rewritten), I've lost on average 25% of the initial capacity per tape. E.g. some tapes started at ~400GB, and are now somewhere at 300GB. some started at 160GB, and are now somewhere around 120GB. Write-speed wise, I also lose speed: The fastest tapes (most often also the tapes which have the highest capacity) started out at 50MB/s, and are now somewhere at 42MB/s. The worst tapes are somewhere between 20MB/s, and are now at 16MB/s. Because most of the tapes were second-hand (we got tape drive, controller, autoloader, and tapes as a gift, as our NGO doesn't have any money), I blamed the bad state of tapes (I verify volume to catalog after each job, I assume I'm still fine. Also there are no tape alerts). But in the course of this year, I also introduced some brand new (never written) tapes. And they are also losing capacity, and write speed to the same amount. Hardware compression, (software) encryption, and software compression is all ON since the very beginning (so no configuration change). And I don't think this can be relevant, because tapes are losing capacity over time, whether they started off at ~406GB (new) or 160GB (used, and initial bad condition). It also doesn't depend what I back up (everything is encrypted anyway, so compression shouldn't work at all at the drive's side). I don't know the logic how bareos recognises that a tape is full, so I thought I give it a try to fill some tapes with zero/urandom to "realign" it. I could fully write to those tapes (=400GB). But bareos continued to only partly use them as before. I then tried to fill the tapes up with btape speed, but the same problem persists when using the tape again in bareos. Tape backups typically have job logs like: *Joblog* Connecting to Director localhost:9101 1000 OK: pavlov-dir Version: 17.2.4 (21 Sep 2017) Enter a period to cancel a command. list joblog jobid=7665 Automatically selected Catalog: MyCatalog Using Catalog "MyCatalog" 2018-08-18 06:37:51 pavlov-dir JobId 7665: Start Backup JobId 7665, Job=edite_backup_automated.2018-08-18_05.30.00_34 2018-08-18 06:37:52 pavlov-dir JobId 7665: Using Device "Ultrium920" to write. 2018-08-18 06:37:53 delaunay-sd JobId 7665: 3307 Issuing autochanger "unload slot 2, drive 0" command. 2018-08-18 06:40:22 delaunay-sd JobId 7665: 3304 Issuing autochanger "load slot 1, drive 0" command. 2018-08-18 06:41:06 delaunay-sd JobId 7665: 3305 Autochanger "load slot 1, drive 0", status is OK. 2018-08-18 06:41:15 delaunay-sd JobId 7665: Volume "XM2705L3" previously written, moving to end of data. 2018-08-18 06:41:33 delaunay-sd JobId 7665: Ready to append to end of Volume "XM2705L3" at file=5. 2018-08-18 06:41:33 del
Re: [bareos-users] losing LTO3 tape capacity with bareos. But full tape length can be written with dd or btape speed
Hi Stefan, yes, I use newly purchased ones when the autoloader/drive requested it. I got clean requests when getting I/O errors with really bad tapes (which I then dumped right away). Otherwise there have been so far no cleaning requests (which is normal to my experience), and the status of the drive is OK. I also looked up regularly logged warnings/errors, and there is nothing (except the aforementioned I/O errors) On Tuesday, August 21, 2018 at 10:38:26 PM UTC+2, Stefan Klatt wrote: > Hi Tim, > > > > only a idea... did you use cleaning tapes? > > > > Regards > > > > Stefan > > > > > Am 21.08.2018 um 15:46 schrieb Tim > Banchi: > > > > Hello, > > I'm using Bareos 17.2 with the following pool, and device configuration: > > Pool { > Name = tape_automated > Pool Type = Backup > Recycle = yes # Bareos can automatically recycle > Volumes > AutoPrune = yes # Prune expired volumes > #Recycle Oldest Volume = yes > RecyclePool = Scratch > Maximum Volume Bytes = 0 > Job Retention = 365 days > Volume Retention = 4 weeks > Volume Use Duration = 12 days > Cleaning Prefix = "CLN" > Catalog Files = yes > Storage = delaunay_HP_G2_Autochanger #virtualfull test > # Next Pool = tape_automated > } > > Device { > Name = "Ultrium920" > Media Type = LTO > Archive Device = /dev/st1 > Autochanger = yes > LabelMedia = no > AutomaticMount = yes > AlwaysOpen = yes > RemovableMedia = yes > Maximum Spool Size = 50G > Spool Directory = /var/lib/bareos/spool > Maximum Block Size = 2097152 > # Maximum Block Size = 4194304 > Maximum Network Buffer Size = 32768 > Maximum File Size = 50G > Alert Command = "/bin/bash -c '/usr/sbin/smartctl -H -l error %c'" > } > > I'm closely monitoring tape performance (write speed + capacity marked as > full by bareos), and all tapes are losing capacity, and write speed, in an > almost linear way. > > Over the period of roughly 8 months (I installed bareos early this year) and > 3 to 4 write cycles per tape (=every 2 months a tape is rewritten), I've lost > on average 25% of the initial capacity per tape. E.g. some tapes started at > ~400GB, and are now somewhere at 300GB. some started at 160GB, and are now > somewhere around 120GB. > > Write-speed wise, I also lose speed: The fastest tapes (most often also the > tapes which have the highest capacity) started out at 50MB/s, and are now > somewhere at 42MB/s. The worst tapes are somewhere between 20MB/s, and are > now at 16MB/s. > > Because most of the tapes were second-hand (we got tape drive, controller, > autoloader, and tapes as a gift, as our NGO doesn't have any money), I blamed > the bad state of tapes (I verify volume to catalog after each job, I assume > I'm still fine. Also there are no tape alerts). > But in the course of this year, I also introduced some brand new (never > written) tapes. And they are also losing capacity, and write speed to the > same amount. > > Hardware compression, (software) encryption, and software compression is all > ON since the very beginning (so no configuration change). And I don't think > this can be relevant, because tapes are losing capacity over time, whether > they started off at ~406GB (new) or 160GB (used, and initial bad condition). > It also doesn't depend what I back up (everything is encrypted anyway, so > compression shouldn't work at all at the drive's side). > > I don't know the logic how bareos recognises that a tape is full, so I > thought I give it a try to fill some tapes with zero/urandom to "realign" it. > I could fully write to those tapes (=400GB). But bareos continued to only > partly use them as before. I then tried to fill the tapes up with btape > speed, but the same problem persists when using the tape again in bareos. > > Tape backups typically have job logs like: > > *Joblog* > Connecting to Director localhost:9101 > 1000 OK: pavlov-dir Version: 17.2.4 (21 Sep 2017) > Enter a period to cancel a command. > list joblog jobid=7665 > Automatically selected Catalog: MyCatalog > Using Catalog "MyCatalog" > 2018-08-18 06:37:51 pavlov-dir JobId 7665: Start Backup JobId 7665, > Job=edite_backup_automated.2018-08-18_05.30.00_34 > 2018-08-18 06:37:52 pavlov-dir JobId 7665: Using Device "Ultrium920" to > write. > 2018-08-18 06:37:53 delaunay-sd JobId 7665: 3307 Issuing autochanger "unload > slot 2, drive 0" command. > 2018-08-18 06:40:22 delaunay-sd JobId 7665: 3304 Issuing autochanger "load > slot 1, drive 0" command. > 2018-08-18 06:41:06 delaunay-sd JobId 7665: 3305 Autochanger "load slot 1, > drive 0", status is OK. > 2018-08-18 06:41:15 delaunay-sd JobId 7665: Volume "XM2705L3" previously > written, moving to end of data. > 2018-08-18 06:41:33 delaunay-sd JobId 7665: Ready to append to end of Volume > "XM2705L3" at file=5. > 2018-08-18 06:41:33 delaunay-sd JobId 7665: Sp