This post is more than 5 years old
1 Rookie
•
97 Posts
1
3232
January 29th, 2015 11:00
Is enabled my SP Write Cache in a VNX2 when a SP fails?
Hi, I've read in some documents and in Article Number:000169648 Version:9 - Which events will cause VNX MCx to disable write caching? that the SP Cache is still enable when a SP fails. Is that true?
| Multi-Core Cache Fault Handling | ||
|---|---|---|
| Use Case | Description | Multicore Cache State |
| Single AC power failure, peer SP still running | AC power is removed from one SPS in a system with multiple SPS (1 or 2 per SP) | Enabled |
| Single SPS power failure, peer SP not running | AC power is removed from one SPS in a system with multiple SPS (1 or 2 per SP) and the peer SP is not present. | Vault dump |
| AC power failure to all SPS | AC power has been removed from all SPS. | Vault dump |
| Single SP over temperature | This SP or the power supply for this SP is reporting an over temperature condition | Enabled |
| Dual SP over temperature | An over temperature condition is reported by all SP and power supplies. | Vault dump |
| DAE 0 (0_0) over temperature | An over temperature condition is reported by one or more power supplies in the first drive enclosure | Vault dump |
| Single SPE/DPE power supply fault | A single power supply is reported as faulted in the SPE. | Enabled |
| Single 60-drive DAE power supply fault | A single power supply in the first DAE is reported as faulted and first enclosure is a 60-drive enclosure | Vault dump |
| Multiple power supply faults | Multiple power supplies are reported as faulted. | Enabled |
| Single fan fault | A single fan has faulted. | Enabled |
| Multiple fan faults | Multiple fans are reported as faulted. | Enabled |
| Degraded vault | A single system drive is offline or has been removed. | Enabled |
| Faulted vault | Multiple system drives are offline or have been removed. | Disabled |
| NDU without power supply firmware upgrade | Array software is being upgraded, but the upgrade does not include a Power Supply firmware upgrade for the SPE, DPE or Vault DAE. | Enabled |
| NDU with power supply firmware upgrade | Array software is being upgraded, which includes new firmware for Power Supplies in the SPE, DPE and/or Vault DAE. | Disabled |
| Single SP fault | A single SP in a dual SP system is removed or rebooted or panics or has a hardware fault | Enabled |
| Single SPS communication fault | One SP is unable to communicate with a single SPS. Other SPS is OK. | Enabled |
| Multiple SPS communication faults | Neither SP is able to communicate with the SPS(s) connected to that SP. | Disabled |
| Single SPS cabling fault | One SPS is incorrectly cabled. | Enabled |
| Multiple SPS cabling fault | Multiple SPS are incorrectly cabled on both SP. | Disabled |
If true... what prevents my cache data to be lost if my second SP fails too (hardware fault)?
Regards,
Thanks in advance
Dario
IPM
No Events found!
brettesinclair
2 Intern
•
715 Posts
1
January 30th, 2015 01:00
My expectation would be that if you lost the second SP suddenly (from catastrophic hw failure), there's a very high probability you will have a data loss/corruption condition, as a cache destage to vault would not occur, and you have no 'clean' cache data on either sp.
LUNs with pending writes in cache would be marked dirty and taken offline requiring you to engage support/engineering to assist. Not fun.
Not a common condition at all though and there's a lot of safeguards to assist in preventing this from happening.
Although, cache protection and recovery has generally improved with each gen release, so maybe that's different again in vnx2, interesting to find out.
Dr.diSousa
1 Rookie
•
97 Posts
0
May 6th, 2015 08:00
Ok, that's what I thought.
Thank you!