HPE Storage Users Group

Hey Guys,

Have anyone of you ever faced issues with write caching ? Have you ever seen a situation where its gets disabled on certain activities like DO etc?

I have observed this in T400 running 3.1.1 version where caching gets disabled.

Others can chime in, but from an architectural standpoint the only thing that should disable WC is if you have a 2-node model and one node is offline, persistent cache on 4 node arrays should mean WC is never disabled so long as the array is online.

That's how I understand it too, it's one thing hp use as a major selling point.

We did not even encounter a single "delayed ack" (we monitor those via srstatcmp - export and cacti/rrd) yet, even under extra heavy load on one of our V400. Let alone a really disabled cache...

Should only happen during a controller updgrade of a dual controller system,
or if something is wrong.

Okay. This is interesting.

We faced this issue where we noticed that cache got disabled and all nodes were up (4 nodes).
The only difference was that we added few disk cages and manual rebalancing was done (tunevv).

While investigating this was the reason given by HP support and they had to enable it. I believe the way to check this is to issue showld command and see if 'WThru' column has 'N' for logical disks.

Wanted to share and see if anyone else has ever faced this issue.

There is no way to manually turn off writeback cache, there are certain operations where write cache could be disabled in order to guarantee data consistency but DO/Tuning operations are not one of those. I seem to remember there was an issue where the flusher state would need to be reset after certain upgrades on older firmware and would need to be manually reset as part of the upgrade. Perhaps that is what you encountered, but rather than speculate further I'd go back to support and request a documented explanation / root cause analysis.