Unsolved
This post is more than 5 years old
10 Posts
0
4653
VPLEX back end path health
In the VPLEX best practices it states that there should be be 2 or 4 (optimal) paths from the back end to a volume. If using 4 paths and one of them goes down then both Unisphere and vplexcli (various health check commands) indicate that there is no issue. It is not until just one path remains that either tool highlight a problem. At this time the number of paths is reduced from 4 to 1!
Has anyone come across this issue and a way to effectively monitor the back end for intial path failure?
Anonymous
5 Practitioner
5 Practitioner
•
274.2K Posts
1
December 4th, 2014 21:00
Not sure how helpful it will be for you ? you could monitor for the number of itls for a specific storage volume in the /clusters/cluster-1/storage-elements/storage-volumes context.
HergenLange
4 Posts
0
March 13th, 2015 07:00
Have written a little python script which monitors all FC Ports (FE , BE , WAN). It uses the REST API for getting the needed Infos. Let me know if you like to test it in your environment.
Just don't know if i can upload it.
Send an email to hergen dot lange at olb dot de
bodenj
10 Posts
0
March 13th, 2015 07:00
Hergen,
I'd be more than happy to have a look at what you have developed. I'll send you an email.
bodenj
10 Posts
0
March 13th, 2015 07:00
Rajesh,
We have over 2000 devices attached to the VPLEX so the ITLs report if not filtered is long. I'll have a look at this though.
TadhgConcannon
1 Rookie
1 Rookie
•
22 Posts
0
March 13th, 2015 07:00
Hi Hergen
I would be interested in looking at this
Sent you an email
Regards,
Tadhg
HergenLange
4 Posts
0
March 13th, 2015 08:00
Hi,
the script only shows you if the FC Ports are online !
Think you want to check if there are a least 2 path to each Storage Volume are online, right ?
bodenj
10 Posts
0
March 13th, 2015 08:00
I logged a call with EMC and they sent a Custom call home configuration. This now causes a call home on the first and subsequent failures.
VPlexcli:/notifications/call-home> view-event-modifications -c
Customer-specific events :
-------------------------
event code name severity
---------- ------------------------------------ --------
0x8a2d905e SCSI_ITL_LOST_LAST_PATH CRITICAL
0x8a2d305d SCSI_ITL_LOST_1_OR_2_PATHS_LEFT WARNING
0x8a2d305c SCSI_ITL_LOST_MORE_THAN_2_PATHS_LEFT WARNING
0x8a2d605f SCSI_IT_LOST_AND_LAST_ITL_PATH ERROR
HergenLange
4 Posts
0
March 13th, 2015 08:00
Hope you are happy with that solution.
bodenj
10 Posts
0
March 13th, 2015 08:00
Hergen,
The vplexcli shows the status of the FC ports. What would be really useful is something similar to powermt display which shows summary counts of active and dead paths to devices.
HergenLange
4 Posts
0
March 13th, 2015 17:00
totally agree.
You are right, but vplexcli is not as comfortable as it should be. You have to login to vplex ; start vplexcli and again give username and password. Not scriptable at all i think.
Crazy for one who knows symcli ;-)
For now the only solution of controlling / monitoring VPLEX from outside is using the REST API, isn't it?
What's powermt display ? Sounds like powerpath. Not using it for my Solaris Servers.
Can you paste some output of it ?
bodenj
10 Posts
0
March 16th, 2015 02:00
powermt is part of the powerpath suite. Example here shows Unix host attached to 56 Symmetrix devices and 2 VPLEX devices. (not a usual configuration as normally you would encapsulate all or nothing) All paths to all devices are ok.
# powermt display
Password:
Symmetrix logical device count=56
VPLEX logical device count=2
==============================================================================
----- Host Bus Adapters --------- ------ I/O Paths ----- ------ Stats ------
### HW Path Summary Total Dead IO/Sec Q-IOs Errors
==============================================================================
2 fscsi2 optimal 60 0 - 0 0
3 fscsi3 optimal 60 0 - 0 0
pkumar50
1 Message
0
June 5th, 2019 03:00