Catalogue not updated

Hi,

Have you observed situations where the catalogue goes for eons w/o being updated although cta-taped shows what seems to be successful writes on the concerned tape? We have a tape whose last update time was this morning at 9.50 and already has 160+ new files but the catalogue shows nothing

output of cta-admin ta ls -v J01386

      "capacity":"20000000000000",
      "occupancy":"8418313825351",
      "lastFseq":"454",
      "full":false,
      "fromCastor":false

but according to cta-taped we are at fSeq 600+

Apr 12 16:03:55 tpm103 cta-taped: LVL="INFO" PID="26877" TID="36074" MSG="File successfully transmitted to drive" thread="TapeWrite" tapeDrive="6,2,3" tapeVid="J01386" mountId="1" vo="xfel" mediaType="3592JE20T" tapePool="xfelfunctest" logicalLibrary="ctajaglib" mountType="ArchiveForUser" vendor="IBM" capacityInBytes="20000000000000" fileId="63753" fileSize="4407521049" fSeq="623" diskURL="root://131.169.50.116:1099/00007573462F632948BCAB747058A2CC0645" readWriteTime="16.313102" checksumingTime="1.506650" waitDataTime="0.003675" waitReportingTime="0.000095" transferTime="17.823522" totalTime="17.823513" dataVolume="4407521049" headerVolume="480" driveTransferSpeedMBps="247.286914" payloadTransferSpeedMBps="247.286887" reconciliationTime="0" LBPMode="LBP_On"

db

ctadb=# select archive_id, disk_file_id, fseq, creationtime  from get_tape_files('J01386') order by 3 desc limit 1;
 archive_id |             disk_file_id             | fseq |      creationtime
------------+--------------------------------------+------+------------------------
       1828 | 00000D66E018A35A4CCCBCD1AF53E64B7D9B |  454 | 2022-04-12 09:53:39+02

thanks,
Mwai

Finally the catalogue was updated after 12 hours. I deduce that it’s part of normal operations!

Hi Mwai,

The TAPE_FILE table is updated immediately when the file is written. However, the version of CTA you are using does not update the statistics on the TAPE table instantaneously. When tapes are written to, they are marked as “dirty”. The external statistics process updates the statistics on all “dirty” tapes, and marks them as “clean”. At CERN, we run this process once an hour. Perhaps you are running it less frequently?

We have improved this in v4.6.2, now the TAPE table is updated immediately when files are written. We still need to run the statistics process periodically to take account of files which may have been deleted.

Regards,

Michael

1 Like

Hi Michael,

this is commit that you talking about?

Thanks,
Tigran.