amdump now seems to think it should include the old test DLE in new backups. The test DLE is not actually included in the backup as it has no information on where to get it from but it still shows up in the status screen from amdump.
When i removed it i ran amadmin delete for the test DLE and then removed it from the disklist. Log has "INFO planner Adding new disk localhost:test" and amdump log has "localhost:test overdue 19105 days for level 0". I removed the tape which the test DLE was on (and some actual backups with it) with 'amrmtape backup --keep-label --cleanup ROT09' and now amadmin find does not see the test DLE anymore and seems to also have resolved amdump. Do i really need to do it this way or should i have done something else to remove the old DLE? Anton "exuvo" Olsson ex...@exuvo.se On 2022-04-20 15:58, Exuvo wrote:
Resolved, had an old test DLE on the tape it was trying to use. I don't understand why it still loaded the tape and did not give any useful error message about it. My testing: If i make a new logs directory the estimate actually runs. Tried moving out all the logs from that day but it still fails which i found weird. Tried moving out one log at a time to find the culprit but it seems that as long as any one of the old logs is there my storage DLE fails. I was only running the storage DLE as that was the only one that failed during the normal backup schedule. I also tried clearing the curinfo file for the DLE. Next i tried running a full backup with all DLEs and now they all showed as MISSING immediately. Which was very interesting. The only common thing for when the DLEs started being missing is that they want to use tape ROT09. Which is the first tape i did a test backup on. I think the actual issue is that it is not finding a usable tape even tho it loads the tape first as if it thinks it is an valid tape to use. Removing the test DLE with amadmin delete made it start working again. Anton "exuvo" Olsson ex...@exuvo.se On 2022-04-19 03:54, Exuvo wrote:The disk was full of old partial dumps from testing of another backup set so i did not want to flush them to any tapes. The logs "driver: adding holding disk 0 dir /dumps/ size 1949288448 chunksize 9007199254740896" indicate that it is finding the dump disk with correct size of 2TB free. Even if the dump drive was full i should be getting "FAILED [can't dump required holdingdisk when no holdingdisk space available ]" not MISSING on the DLE. I even tested with the holding disk disabled and it still says it is missing. The path is most certainly not missing from the filesystem. I also tried changing to another file path but still MISSING. It must be something in the database that got stuck, i will have to parse it myself and see if i can figure out what. Anton "exuvo" Olsson ex...@exuvo.se On 2022-04-18 02:34, John Lauro wrote:I normally use amflush to clear the holding disk. Not saying rm doesn't work, but that will not update the Amanda info database. It might think that space is still being used. On Sun, Apr 17, 2022 at 2:09 PM Exuvo <ex...@exuvo.se> wrote: I take it you did not read what i wrote. The holding disk is already empty after i 'rm -r':ed it. But even if i tell amanda to never use the holding disk for the failing DLE it just stops immediately when trying to run the planner (it normally takes a few minutes for that DLE). All other DLEs run fine.
