Unsolved
This post is more than 5 years old
2 Intern
•
138 Posts
0
361
March 19th, 2009 06:00
Compact task taking days to complete instead of hours.
Hello,
DX 6.3 for Windows with Centera Storage.
I am performing a compact task on a media folder in order to relocate the data to a SAN location. This usually takes a few hours to compact a media folder of about 150GB in size. I have a task that has been running for 12 hours, the number of files processed and fetched are creaping up by the bytes. For example in the last 10 minutes. about 400 bytes have been transfer.
I've disabled Macaffe on this machine just to rule it out. I also restarted the server and started over (12 hours ago) but the symptom still exists.
I've opened a case with EMC but I wanted to see if anyone had any other suggestions that could help me quickly diagnose.
DX 6.3 for Windows with Centera Storage.
I am performing a compact task on a media folder in order to relocate the data to a SAN location. This usually takes a few hours to compact a media folder of about 150GB in size. I have a task that has been running for 12 hours, the number of files processed and fetched are creaping up by the bytes. For example in the last 10 minutes. about 400 bytes have been transfer.
I've disabled Macaffe on this machine just to rule it out. I also restarted the server and started over (12 hours ago) but the symptom still exists.
I've opened a case with EMC but I wanted to see if anyone had any other suggestions that could help me quickly diagnose.
No Events found!
birthslj1
2 Posts
0
May 6th, 2009 04:00
I was wondering if you had got a resolution to this problem as we are exeriencing the same one. Have go a call open with EMC, but no obvious solutions from there as yet.
Using Windows 2003 SP2, DX 6.10.040, (patch level 09) separate media server, fibre attached Storagetek tape library, moving extended drive from a 2 to a 15Tb LUN (new Dell/EMC Clarion CX4)
File Fetch reasonable performance, but 200GB tape compaction halted after 7 days with 25% processed. Havent used compaction before (data very static) so dont have a benchmark to compare to.
birthslj1
2 Posts
0
May 15th, 2009 09:00
I tried a second tape with similar very slow results.
In an effort to establish exactly what was happening I turned on detailled logging using Services - Event logging - Trace - server processes - "Compact - Tape Requests"
From this i established that after a tape scan (took about two hours for a 200GB tape), files were being retrieved. This is an extract from the log.
MSG 08/05 16:54:55 (08AC) Compact media 000636(624) - Examining file
H:Echosounder Transducer Depth_20040821_rev.xls.
MSG 08/05 16:55:46 (08AC) Compact media 000636(624) - Examining file
H:Echosounder Transducer Depth_20040824_rev.xls.
MSG 08/05 16:56:38 (08AC) Compact media 000636(624) - Examining file
H:\Echosounder Transducer Depth_20040731.xls.
It is taking approx a minute to retrieve a file. This seems to be becasue the files are not retrieved sequentially, instead DiskXtender is using seperate retrieval tasks for each file so the library does a rewind - seek - read - rewind for each file.. Becasue the average file size is very small, there are lots of them on a tape, The retrieval time is similar for all sizes so the the time bottleneck is the tape rewind-seek times.
Knowing the bottleneck does not solve the problem - We have a 2Tb extended volume that has 455,000 files on it. At one minute to retrieve each file, it will take eight months for this to be retrieved using a single tape drive. Our bigger problem is a 15TB extended volume that also has to come back (10years!!)