HPE GreenLake Administration
- Community Home
- >
- Servers and Operating Systems
- >
- Operating Systems
- >
- Operating System - Linux
- >
- How can I gather performance history for redhat li...
Operating System - Linux
1829101
Members
2384
Online
109986
Solutions
Forums
Categories
Company
Local Language
back
Forums
Discussions
Forums
- Data Protection and Retention
- Entry Storage Systems
- Legacy
- Midrange and Enterprise Storage
- Storage Networking
- HPE Nimble Storage
Discussions
Forums
Discussions
Discussions
Discussions
Forums
Discussions
back
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
- BladeSystem Infrastructure and Application Solutions
- Appliance Servers
- Alpha Servers
- BackOffice Products
- Internet Products
- HPE 9000 and HPE e3000 Servers
- Networking
- Netservers
- Secure OS Software for Linux
- Server Management (Insight Manager 7)
- Windows Server 2003
- Operating System - Tru64 Unix
- ProLiant Deployment and Provisioning
- Linux-Based Community / Regional
- Microsoft System Center Integration
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Blogs
Information
Community
Resources
Community Language
Language
Forums
Blogs
Go to solution
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
06-11-2009 08:12 AM
06-11-2009 08:12 AM
Re: How can I gather performance history for redhat linux
When you say I/O reporting sucks, are you specifically talking about device mapper naming or do you have somethign else in mind.
As for DM naming, I think I follow what you're doing but I'm not sure where you're getting names like "asm-db, asm-dbp1, asm-fra, asm-frap1, etc".
Perhaps I can add what you're looking for to collectl if not too tough and general enough...
-mark
As for DM naming, I think I follow what you're doing but I'm not sure where you're getting names like "asm-db, asm-dbp1, asm-fra, asm-frap1, etc".
Perhaps I can add what you're looking for to collectl if not too tough and general enough...
-mark
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
06-11-2009 01:29 PM
06-11-2009 01:29 PM
Re: How can I gather performance history for redhat linux
Well, the suck statement was really referring to the default stats collection that comes with Zabbix: a small blot on an otherwise superb product. The rest of what I said though referred to my frustration with tools in general that report stats using dm device names, and my work around to get something more meaningful.
The names "asm-db, asm-dbp1, asm-fra, asm-frap1, etc" come from multipathd. By default it aggregates device names that present different paths to the same device (/dev/sdc, /dev/sdd, etc) into a single device name e.g: /dev/mapper/mpath0. The fancy names come from multipathd's user friendly names feature, where you bind the WWID of a device you know, to a name you want to use and declare it in /etc/multipath.conf.
So for example if I know (from the LUN and from using the command "multipath -ll") that /dev/mapper/mpath0 is a disk I'm using for Oracle ASM data, and that it has a WWID of 16465616462656166353a3500000000000000000000000000 then I can rename mpath0 to something more meaningful by adding this to multipath.conf
multipaths {
multipath {
wwid 16465616462656166353a3500000000000000000000000000
alias asm-db
}
}
Then /dev/mapper/mpath0 becomes /dev/mapper/asm-db. I can treat that device then just like I would a regular disk block device, with the added benefit that the new name appears in a "multipath -ll" listing. Unfortunately it doesn't appear in /proc/diskstats, so I have to find its associated dm device first.
Does that help?
John
The names "asm-db, asm-dbp1, asm-fra, asm-frap1, etc" come from multipathd. By default it aggregates device names that present different paths to the same device (/dev/sdc, /dev/sdd, etc) into a single device name e.g: /dev/mapper/mpath0. The fancy names come from multipathd's user friendly names feature, where you bind the WWID of a device you know, to a name you want to use and declare it in /etc/multipath.conf.
So for example if I know (from the LUN and from using the command "multipath -ll") that /dev/mapper/mpath0 is a disk I'm using for Oracle ASM data, and that it has a WWID of 16465616462656166353a3500000000000000000000000000 then I can rename mpath0 to something more meaningful by adding this to multipath.conf
multipaths {
multipath {
wwid 16465616462656166353a3500000000000000000000000000
alias asm-db
}
}
Then /dev/mapper/mpath0 becomes /dev/mapper/asm-db. I can treat that device then just like I would a regular disk block device, with the added benefit that the new name appears in a "multipath -ll" listing. Unfortunately it doesn't appear in /proc/diskstats, so I have to find its associated dm device first.
Does that help?
John
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
06-11-2009 01:39 PM
06-11-2009 01:39 PM
Re: How can I gather performance history for redhat linux
So if I understand you correctly, I can add my own aliases to multipath.conf - something I've never done before, but that doesn't say I can't try. When I look at my conf file most of the entries are commented out, including the one for multipaths. Is the wwid listed there the real one? Do I just uncomment and add an alias as you did? Here's what's in that stanza on my system:
#multipaths {
# multipath {
# wwid 3600508b4000156d700012000000b0000
# alias yellow
# path_grouping_policy multibus
# path_checker readsector0
# path_selector "round-robin 0"
# failback manual
# rr_weight priorities
# no_path_retry 5
# }
# multipath {
# wwid 1DEC_____321816758474
# alias red
# }
#}
tell me what it should look like and where the alias will appear and I'll see what I can do, maybe even get it into the next release of collectl!
-mark
#multipaths {
# multipath {
# wwid 3600508b4000156d700012000000b0000
# alias yellow
# path_grouping_policy multibus
# path_checker readsector0
# path_selector "round-robin 0"
# failback manual
# rr_weight priorities
# no_path_retry 5
# }
# multipath {
# wwid 1DEC_____321816758474
# alias red
# }
#}
tell me what it should look like and where the alias will appear and I'll see what I can do, maybe even get it into the next release of collectl!
-mark
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
06-11-2009 02:31 PM
06-11-2009 02:31 PM
Re: How can I gather performance history for redhat linux
Ah, well you can't use this for direct attached disks that just hang off a local scsi/sata bus and only have one path to them. Multipathd can only be used to manage disks that have multiple IO paths. Like for example if you had two FibreChannel HBA cards linked via dual FC Switches to the 4 ports of an EVA over a SAN, in which case you would get 4 physical paths to each device. Or another example, where you're presenting a disk from one system to another via iSCSI, and you have dual NICs and present the disk over both of them, in which case you'd get 2 paths.
Tell you what, forget about multipathd for a minute and look at LVM, because this applies to LVM Logical Volumes just as well as multipath devices, and that's probably going to be more accessible to you without dedicated hardware.
On my systems I like to name what would normally be the default LVM volume group: VolGroup00 to be "system" and the two LVs that sit on that as "root" and "swap". device-mapper creates the following two devices for me:
/dev/mapper/system-root
/dev/mapper/system-swap
I think Ubuntu does this by default anyway, but I mostly use Redhat or CentOS.
But these devices don't appear in /proc/diskstats either. So I treat them just like I would user friendly multipath device names. I pick them out of /dev/mapper, and use the major/minor number pair to find the dm device files and then record the disk stats against system-root instead of dm-0 and system-swap instead of dm-1. If you can solve that, then you'll automatically solve the multipath user friendly names scenario as well :)
John
Tell you what, forget about multipathd for a minute and look at LVM, because this applies to LVM Logical Volumes just as well as multipath devices, and that's probably going to be more accessible to you without dedicated hardware.
On my systems I like to name what would normally be the default LVM volume group: VolGroup00 to be "system" and the two LVs that sit on that as "root" and "swap". device-mapper creates the following two devices for me:
/dev/mapper/system-root
/dev/mapper/system-swap
I think Ubuntu does this by default anyway, but I mostly use Redhat or CentOS.
But these devices don't appear in /proc/diskstats either. So I treat them just like I would user friendly multipath device names. I pick them out of /dev/mapper, and use the major/minor number pair to find the dm device files and then record the disk stats against system-root instead of dm-0 and system-swap instead of dm-1. If you can solve that, then you'll automatically solve the multipath user friendly names scenario as well :)
John
- « Previous
-
- 1
- 2
- Next »
The opinions expressed above are the personal opinions of the authors, not of Hewlett Packard Enterprise. By using this site, you accept the Terms of Use and Rules of Participation.
Company
Events and news
Customer resources
© Copyright 2025 Hewlett Packard Enterprise Development LP