β06-26-2013 06:53 AM
Hi all,
So on a support case the topic of failed items in indexes has come up. Out of a user population of 8000 we had around 600 archives with indexes with failed items in them. I think this is just because we have never tackled this before.
So I ran the below SQL query to fetch them all and I will work through them in the coming days.
USE EnterpriseVaultDirectory
SELECT IV.FailedItems AS 'IVFailedItems', IV.FolderName AS 'IV.FolderName', EME.MbxDisplayName AS 'EME.MbxDisplayName'
FROM EnterpriseVaultDirectory.dbo.IndexVolume AS IV, EnterpriseVaultDirectory.dbo.ExchangeMailboxEntry AS EME
WHERE IV.FolderName = EME.DefaultVaultId
AND IV.FailedItems is not null
AND IV.FailedItems > '0'
GROUP BY IV.FolderName, IV.FailedItems, EME.MbxDisplayName
ORDER BY EME.MbxDisplayName
I guess my question is: Does anyone thing this is a significant issue? Or does anyone know of any issues which are caused by this?
Thanks,
Jeremy.
Solved! Go to Solution.
β06-26-2013 07:53 AM
Well we're not talking failed indexes are we
we're talking missing items from the index, which in the database is logged as Failed Items
which typically means that EV couldn't add those items to the index, which usually means that storage couldn't get to the DVS at the time of indexing for whatever reason
if it was that the indexable item was corrupt or empty, then it would be Missing Content which isn't logged in the index
If you look at the reason in the missingitems.txt file, you'll have a reason code there, usually its -1, meaning unable to retrieve from storage
i'll see if i can find you a script that will automate the repair process for you,
β06-26-2013 06:56 AM
You don't mention the version or history of versions ... is this a nice clear 10.0.3 CHF2 fresh install, or has it been around a long time? Where are the indexes stored?
β06-26-2013 07:04 AM
Hi,
We are currently on version 8.0.5, upgraded from 7.5, upgraded from 7.0.
Indexes are stored on a dedicated drive in each of our 7 EV email archiving servers.
Thanks,
Jeremy.
β06-26-2013 07:06 AM
Failed items are fairly easy but time consuming when you have 600 archives with missing items
all you need to do in EV9 is perform a repair and it will reindex the items in the itemsmissing.log file
if you have EV10 you can just do a synchronize
Don't rebuild the archives though, simple repair should be enough
β06-26-2013 07:07 AM
I guess it's not *super* unusual to have some indexes showing as failed, given the history, though you do have what sounds like stable hard drives (rather than some rough old NAS).
Is there any pattern? 10% .. is it across all EV servers, or just one or a few? Those archives have anything in common?
What have support advised you to do?
β06-26-2013 07:17 AM
Hi guys,
So the issue I have is these alerts coming up in the VAC:
Deleted items waiting to be deleted from indexes | How to fix | |
There are 11713 items that have been deleted from Vault Store 'MailboxVS02' but have not yet been deleted from the indexes. | ||
Example symptoms: users who search for items may find items that have been deleted, leading to calls to your Help Desk. Deleted items can still be found by Discovery Accelerator and Compliance Accelerator. There may be a problem with the Indexing service. |
It comes up for multiple servers. The archives effected do not have anything in common, all on different servers.
Support have suggested the failed items in indexes could be a cause or at least part of the issue.
The drives are all SAN based, our EV and Exchange environments are all on VMWare. So yes to best of my knowledge we are talking stable, expensive infrastructure.
I have already targetted the ones with the most failed items in them, and I am now working my way through the list of 400-500 with less than 10 failed items.
I saw on another thread someone else appears to be having a similar issue, though on version 9, so I just restarted the Admin service on one box, but select count(*) from journaldelete where indexcommitted = 0 still gives the same figure as before I even started all these index repairs earlier in the day. I have done roughly 200 with a colleague throughout the day today.
β06-26-2013 07:53 AM
Well we're not talking failed indexes are we
we're talking missing items from the index, which in the database is logged as Failed Items
which typically means that EV couldn't add those items to the index, which usually means that storage couldn't get to the DVS at the time of indexing for whatever reason
if it was that the indexable item was corrupt or empty, then it would be Missing Content which isn't logged in the index
If you look at the reason in the missingitems.txt file, you'll have a reason code there, usually its -1, meaning unable to retrieve from storage
i'll see if i can find you a script that will automate the repair process for you,
β06-26-2013 07:56 AM
I know that multiple-index-rebuild is available:
http://thingsilearnedtoday.net/2010/09/06/rebuilding-multiple-indices/
Not sure about repair.