1752810
Members
5668
Online
108789
Solutions
SOLVED
Do you have sufficient swap? What's the output of "swapinfo -tam"
HTH
Duncan
I am an HPE Employee
HTH
Duncan
I am an HPE Employee
You have a 32GB system with only 4GB of swap.
As an absolute minimum you should have at least 25% (8GB in your case) of memory size available as swap. In your case I'd look at going right up to 50%/16GB of swap.
See the following docs for more guidelines:
http://docs.hp.com/en/B2355-90950/ch06s03.html
HTH
Duncan
I am an HPE Employee
As an absolute minimum you should have at least 25% (8GB in your case) of memory size available as swap. In your case I'd look at going right up to 50%/16GB of swap.
See the following docs for more guidelines:
http://docs.hp.com/en/B2355-90950/ch06s03.html
HTH
Duncan
I am an HPE Employee
Well swap has to be contiguous, so I doubt you'll be able to extend that existing swap partition. I'd just create another lvol of 12GB in size and use that - remember these swap areas likely won't ever get *used*, the space just gets *reserved* in case it needs to be used.
HTH
Duncan
I am an HPE Employee
HTH
Duncan
I am an HPE Employee
SEP,
>> Putting swap in local storage is a bad
>> idea.
Is that waht you meant to say, as the rest of your post seems to indicate otherwise?
>> If this system is new, you might want to
>> reinstall the OS properly and avoid these
>> issues.
Agreed - that would be the best policy, but this looks like a prod system - presumably if it is, that sort of outage will have to be planned. Adding a seperate swap volume in the short term will resolve that.
>> Yes, you can probably put the additional
>> 12 GB of space on external storage,
>> however vg00 might not have enough PE's
>> left to allocate space on an external
>> disk.
Well it doesn't all have to go in vg00 - you can have swap in seperate VGs, just so long as you have *some* in vg00 so you can get into single user/LVM maintenance mode.
>> The problem with putting swap on external
>> storage is that if there is an unplug of
>> the fiber network your system halts due
>> to loss of access to swap space.
>> Not a good situation at all.
Unplug a fiber cable on a system with only data on the SAN, and to all intents and purposes the system is hung anyway - after all we don't run systems just for the OS, we run them for the apps! Anyway, this just leads us back in to the old "boot from SAN" argument again, and I know we'll just have to agree to disagree on that one.
Besides, presumably Nirukshitha has at least 2 connections to his SAN?
Nirukshitha, to summarise:
i) In the long run SEP is correct - get all your swap on local disks if you're booting off local disks, at least if you lose the SAN, you can still get to an OS prompt. The simplest way to do this (assuming the disks in vg00 are big enough) is using an ignite backup/interactive recovery where you'll get the chance to change the size of swap.
ii) In the short term to prevent the erors you're seeing add some external swap - it can go in a seperate VG, just make sure you assign it a lower priority than your primary swap volume on vg00. That way it won't be used until primary swap is full.
HTH
Duncan
I am an HPE Employee
>> Putting swap in local storage is a bad
>> idea.
Is that waht you meant to say, as the rest of your post seems to indicate otherwise?
>> If this system is new, you might want to
>> reinstall the OS properly and avoid these
>> issues.
Agreed - that would be the best policy, but this looks like a prod system - presumably if it is, that sort of outage will have to be planned. Adding a seperate swap volume in the short term will resolve that.
>> Yes, you can probably put the additional
>> 12 GB of space on external storage,
>> however vg00 might not have enough PE's
>> left to allocate space on an external
>> disk.
Well it doesn't all have to go in vg00 - you can have swap in seperate VGs, just so long as you have *some* in vg00 so you can get into single user/LVM maintenance mode.
>> The problem with putting swap on external
>> storage is that if there is an unplug of
>> the fiber network your system halts due
>> to loss of access to swap space.
>> Not a good situation at all.
Unplug a fiber cable on a system with only data on the SAN, and to all intents and purposes the system is hung anyway - after all we don't run systems just for the OS, we run them for the apps! Anyway, this just leads us back in to the old "boot from SAN" argument again, and I know we'll just have to agree to disagree on that one.
Besides, presumably Nirukshitha has at least 2 connections to his SAN?
Nirukshitha, to summarise:
i) In the long run SEP is correct - get all your swap on local disks if you're booting off local disks, at least if you lose the SAN, you can still get to an OS prompt. The simplest way to do this (assuming the disks in vg00 are big enough) is using an ignite backup/interactive recovery where you'll get the chance to change the size of swap.
ii) In the short term to prevent the erors you're seeing add some external swap - it can go in a seperate VG, just make sure you assign it a lower priority than your primary swap volume on vg00. That way it won't be used until primary swap is full.
HTH
Duncan
I am an HPE Employee