Monitor Index repair

[ZeNick]

Hello,

I launched the reparation of one of the archive journal index 2 weeks ago and today, it's still running.
That does not surprise me.

But I'd like to know if it exists any tool, log, or anything else that could give me any information about how long it will be running?

Many thanks for your help,
Regards

[JesusWept3]

You did a Rebuild, Repair or Update?

This is important because a Repair actually goes off of the existance of the IndexMissing.log file in the Index location itself
so if an item couldn't be indexed or something like that, EV marks it down in an IndexMissing.log file and then marks down in SQL that an item is missing
when you hit repair, it then opens up the log file and then tries to add each item in the log file back in to the index.
However if that indexmissing.log file itself doesn't exist but you ran a repair, then what would happen is it would sit at rebuilding and do nothing, would never finish

That being said, if you run a rebuild, it will delete the index entirely and then go throough and rebuild the index one item at a time which can take a long long long time
the updates will just go through the Highest index sequence number in the Index meta data and compare to the highest index sequence number in the database

so if the database says it has a highestISN of  5000, but the index metadata says 4000 then when you hit update, it will start indexing item 4001, 4002, 4003 etc all the way until completion.

When an update or a rebuild occurs, you should have an event every so often that says the index is a certain percent complete and should give an ETA of when it will finish rebuilding.

You can also look in the index location itself for a file called updates.log which will show how many items have been added at at least 15 minute intervals

What i would suggest other than this is just start a dtrace against the indexserver, restart the indexing service and then perform a repair again, and then look in the log file and see if you see any activity or errors.

[ZeNick]

Hello,

Thank you for your answer.

I launched a "repair"
In fact, I launched 2 "repairs", the fist one solved about 30 items on 100 failed. I then lauched a second one wich ran for 2 weeks.

I stopped the process last friday as there wasn't any file modified in the concerned index folder since the repair was launched and I restart a repair that doesn't seems to do anything.
I checked in the index folder and there is no indexmissing.log file, maybe the problem is there?

I also launched a dtrace on indexserver but there are so many information that I'm not able to get the relevant ones. Do you know what should I search for?

Do you have any suggestions?

Many thanks for your help

[JesusWept3]

OK so firstly lets make sure you're looking at the right index locations.


1. Run the following query and in Windows Explorer go to the one that is rebuilding

SELECT A.ArchiveName, (IRP.IndexRootPath + '\'+ IV.FolderName) AS Folder, IV.FirstItemSequenceNumber AS FirstISN, IV.HighestItemSequenceNumber AS LastISN, IV.IndexedItems, IV.Rebuilding, IV.Failed, IV.FailedItems
FROM IndexRootPathEntry IRP,
       IndexVolume IV,
       Archive A,
       Root R
WHERE A.RootIdentity = R.Rootidentity
  AND R.RootIdentity = IV.RootIdentity
  AND IV.IndexRootPathEntryId = IRP.IndexRootPathEntryId
  AND A.ArchiveName = 'your Archive Name'

2.  Run the following command in a command prompt in your Enterprise Vault directory, and use the Folder given from the query above
indexcheck.exe –c MissingDocs –f <yourIndexFolder> -db yourSQLServer

Once that has completed, it will have regenerated the IndexMissing.log file, and you can run the repair again which should complete fairly quickly

As for the other index that only repaired 30/100 items, you need to look at the event viewer as it will throw an event for each item that couldnt be indexed most likely because it could not be found on disk

[ZeNick]

Once again, thank you for your help.

I'm sorry but I finally decided to launch a complete rebuild of the index.
I 'm sure that your solution would have been ok but I wanted to restart with a clean index.

The rebuild will take about 10 days but I can monitor its status.

Anyway, I'm interested with your solution and will try it on other indexes.

Regards