AnsweredAssumed Answered

Rebuild Indexes keeps running a server out of memory

Question asked by dougbrowning on Aug 10, 2015

I have a standard weekly maintenance script that runs Compress, Rebuild Indexes, and Analyze each weekend (python script called from a bat file).  It runs fine on 9 of our 10 servers (slight ram increase but not so much it is an issue).  But 1 server runs up the memory to the high 90% range then either gives a out of memory error and ends the script or it completes the script but leaves so little memory left that I have to reboot the server after.  When I run it manually I can see the ram steadily creep up and up - 2-10% increase for each db it hits - there are just 5 sde dbs.  A fresh reboot will run ram from 25% to about 75%.  At the end of the week the server is at about 55% and runs to 95-98% then gets mad.

 

Most Dbs the rebuild takes 1-4 minutes but I have one db that takes 13 minutes - and this db seems to grow the ram the most.

 

I have tried everything I can think of but it still happens every single weekend.  It has replica copies of the same DBs that other servers have so not sure why this one server has an issue.  The rebuilds take about the same about of time as other servers.

 

I did find several bad spatial indexes (9999999 to -9999999 on the extent) and fixed them all but still does it.  Have checked all regular extents and all good.

 

I then set the max ram in sql but it still goes right past it and runs out.

 

This server is part of a VM that had log issues once that affected memory but that has been all cleaned up.

 

I do see that it lists the GeoDB version as 10.1 where we are 10.2.2.

 

In the Domain, Resolution, Tolerance tab I have a min Y of -9998099.99992653 and a Mix X of -5120899.99993323 which seems odd.  This is a UTM 10 db.

 

Any other ideas on what I can try?

 

 

Win 2008 R2 Server Enterprise 64 bit

Arc 10.2.2

SDE MSSQL 2008 R2

 

Thanks

Outcomes