I have a problem very similar to this guy
http://www.ibm.com/developerworks/forums/thread.jspa?threadID=204477
and I replied there but thought I should probably start a topic also.
A few months ago we put in a x3650 with megaraid and an exp3000 box. The total is 2TB. Occasionally and with increasing frequency we cannot access the data on the exp3000. On the 3650 I can alt tab, browse windows, ctrl alt delete, no cpu or memory usage. As soon as I hit something like my computer or network places which would show the data, (local exp3000 is D network users see U) that window will just hang until the exp3000 answers. I've talked with IBM and Microsoft with no luck so far. IBM had me do updates and Microsoft wants me to run a chkdsk which I can't until this coming weekend. On two TB that is going to take forever.
Here you are. Quick update on what I've done lately. I THINK I'm completely up to date now on firmwares and such. I updated the firmware to the exp3000 last night. I'm sure the dsa will tell you but we have a serveraid mr10, x3650 and exp3000 box.
Many thanks for your time!
Updated all those and will be rebooting at 5. Want a dsa after that? I had wondered about the NIC driver update possibly dropping network and sure enough it did. It said I didn't have to reboot now but I lost that nic until I reboot apparently. I already had the second nic programmed with IPs though because I was going to work on possibly doing some load balancing so a quick IP change and it was back.
Worth noting though if you ever recommend someone update it. None of the others change anything til reboot.
I can post a fresh DSA afterwards.
I studded a case with poor disk performance with EXP3000 with Megaraid 8480E.
The main problems was bottleneck in write operation. Client had to update Raid controller firmware, driver version and install Windows 2003 updates, to get possibility to enable in Windows write back operations. Before this Windows did only write through operations, which made disk IO very slow when write operations got intensive.
There was another issue specific related to Megaraid. It uses 128MB memory for specific internal requirement, and only 128MB for data cache.
So far anyway the problem seems solved. I hesitate to say that for sure because of how random the event is. However, for whatever reason thursdays and fridays always seemed to be the worst and we had no hiccups last week.
The problem is back as of 1:30 ish today. The main problem seems to be disk WRITE time. This is causing disk que etc to be very high. How can I see what is writing so much? Don't forget this is a raid 6 which can affect that but GEEZ this much?
THanks any thoughts are appreciated.
No actually the option to enable write caching is available to check and once I do so it also makes the next option available which is the advanced performance.
But both are prempted by "this device does not allow its write cache setting to be modified" so the setting does not stick. The gentleman in the post I reference above did firmware updates like myself and the final blow was enabling write caching. His card is not the same as mine however.
Thanks for all your help so far.
