3par for host port path dropping

[koopa79]

Guys,

I'm looking for a way of being notified when the 3par loses a path to a Host.

I can see the following command would let me know that the port has dropped in the last 10 minutes

showeventlog -debug -oneline -min 10 -msg "Host Path Status Change"

The question is, anyone have an idea i can get this running as a script and then email if it finds anything with that criteria? As you may have guessed i'm a script man.

Thanks

[apol]

Hi,

basically you "just" need to run your command every 10...60...xxx minutes and look if the returned data contains relevant data.

We thought about this as well, but as a simple esx reboot would trigger a relevant amount of alerts we decided to not do it. You could of course check if after an appropriate amount of time you see the corresponding "Server path is back"-messages, but you get a lot of those as well when you reboot a host.

[koopa79]

apol i understand i need to run this every 10 mins etc, which if we knew we were going to reboot a ESX host then we could disable it briefly. As i was thinking of having this a scheduled task running every 10 mins or so.

So any idea how i can make this a simple script, would people make a powershell script to run this command? or any other way of having this run?

I would rather get alerts for a known reboot of a host than not be aware of a path dropping but not being alerted as 1 or more paths are still active.

Thanks

[ailean]

I use HPE/Brocade SAN Network Advisor for this, although it's only looking at the active ports rather than paths. I did have email alerts set at one point but with 130+ servers across 4 teams it was just a pain with all the unannounced reboots here.

It still logs the alerts and I can spot them in the log window but I also have a report set to check every server has two online ports in the morning so I can see a quick summary.

Occasionally I'll sort the 3PAR host list by Storage System Ports to see if there are any odd numbers and also do that before any 3PAR or SAN upgrades just for peace of mind but it's been a long while since I've seen a glitch path wise that wasn't an offline port.

I don't even get told when people decommission servers, just have to poke them when something has been down for a while to find out if it's coming back.

[koopa79]

Ailean,

I did look previously at using the DCNM software to report on this, as it does even have a selection for showing what doesn't have redundant paths.

The only problem is we only have 2 alias per zone we create, so if a server has 2 HBA's and we zone up each HBA to 2 different ports on the 3par, we actually create 4 zones to do this, where as i expect most people would zone 1 HBA to 2 x 3par ports in 1 zone.

I believe previously it was best practice to have a 1 to 1 zone, but nowadays its seems to be 1 HBA to as many 3par ports in 1 go.

This means this report thinks we don't have redundant paths, which we do, so the report from DCNM is no good.

[ailean]

I use 1-1 zones too, still think it's the recommended (for 3PAR) but they added the 1-many storage ports because of limitations with other HPE storage (maybe XP?), that had to have all the storage ports of all arrays in a zone with 1 host port I believe.

It's the Check for multiple fabric connections with a min of 2 that I use within the Host Checks section of BNA, as mentioned it's only checking the ports are connected not paths/zones so doesn't cover everything.

Something at the 3PAR end would be handy, i.e. if the known valid values are either 4 paths or 0 paths then alert on any other values that last for more then 5/10 mins.

[apol]

koopa79, if you check the logs for messages in the last 10 minutes, you have to do that every 10 minutes if you want to cover 100% of a day...

You could write a script that checks this once a hour, once a day,... In that case, you could try to work with the output of showhost instead of the log file: You don't care if a host was rebooted once, twice, ..., you just want to check that every host has all paths at the beginning of the work day, at a time when there are no planned reboots.

If you use cli.exe or the powershell cmdlets doesn't matter imho, it's just one command with a lot of script around it to parse the output.

[koopa79]

apol,

yes the idea was to have scheduled task running every 10 mins to check for that error, and if it sees it to drop us a mail.

as anyone got a basic script like that i can amend that would be handy, as that's the part i don't know how to do?

Thanks

[kk2491]

Hi koopa79,

You can create a script using plink connection. I have developed a similar codes/scripts for health check, capacity report etc etc.

Let me check and msg you the relevant code.

Regards,
KK

[koopa79]

Hi koopa79,

You can create a script using plink connection. I have developed a similar codes/scripts for health check, capacity report etc etc.

Let me check and msg you the relevant code.

Regards,
KK

If you have some scripts that i can look at, that would be great. I tried to message you directly through this, but i couldn't get it to work.