This post is more than 5 years old
51 Posts
0
2686
June 22nd, 2009 12:00
SPS and write cache
We had a issue where a Dr array that we mirror too had a failed SPS. This produced Alerts stating that the write cache was disabled. Prior to replacing the SPS we had serveral remote mirrors go in and out of sync. I suspected that this would happen but i was told by our supplier that the write cache does not disable.
I ran some Cli against the SP and infact it said that the cache remained enabled.
I am a little confused and would like to hear other points of view
I ran some Cli against the SP and infact it said that the cache remained enabled.
I am a little confused and would like to hear other points of view
No Events found!


kelleg
6 Operator
•
4.5K Posts
0
June 23rd, 2009 12:00
glen
DGM3
238 Posts
0
June 22nd, 2009 13:00
Thanks,
DGM
sidanat
51 Posts
0
June 22nd, 2009 13:00
DGM3
238 Posts
0
June 22nd, 2009 14:00
That said, when someone goes to physically replace the failed SPS unit - the act of disconnecting the it from the array is going to temporarily result in a power supply in the CX700's base enclosures (where the Storage Processors and first tray of drives are) losing AC input - that will disable write cache, until the new SPS unit is connected (thus restoring AC input).
Or said another way - on a CX700 a single SPS unit failing-to-hold charge won't cause write cache to disable, but interrupting AC input (such as during maintenance activity) to a power supply "behind" an SPS unit will cause write cache to disable.
Does that info help explain what was observed?
Regards,
DGM
sidanat
51 Posts
0
June 23rd, 2009 02:00
Looking at the alerts we received it seems that the array considered that the SPS had been removed therefore the write cache was disabled. See below
"Device Enclosure 0 SPS A
Description The UPS failed or the UPS serial cable to the storage system is not connected. Write cache was disabled and will remain disabled until you repair the UPS. See the AX-series documentation for replacement instructions."
That would then fit in as far as the mirrors going in and out of sync ?
but the odd thing when i ran the naviseccli getall -sp this reported that the R/W cache was indeed enabled.
So i was getting conflicting output from the Array?