c6, LSI megaraid drive failure notification

List overview All Threads
Download

newer

older

Kernel panic after install in...

Serial port driver on CentOS 6

John R Pierce

22 Dec 2011 22 Dec '11

9:12 p.m.

i'm configuring a storage server with CentOS 6.2, it uses a LSI MegaRAID SAS controller, I'm using LSI's megacli to configure the storage... Any ideas on how to get drive failure notifications out of this system? I'm configuring hot spares but I'd still like some sort of notification when a drive has failed so the spare can be replaced.

-- john r pierce N 37, W 122 santa cruz ca mid-left coast

Show replies by date

Craig White

22 Dec 22 Dec

10:32 p.m.

On Dec 22, 2011, at 1:12 PM, John R Pierce wrote:

...

i'm configuring a storage server with CentOS 6.2, it uses a LSI MegaRAID SAS controller, I'm using LSI's megacli to configure the storage... Any ideas on how to get drive failure notifications out of this system? I'm configuring hot spares but I'd still like some sort of notification when a drive has failed so the spare can be replaced.

---- don't know how to do it on CentOS but on Ubuntu, I use megaclisas-status package which goes hand in hand with megacli and it sends notifications.

If you want, I can e-mail you the megaclisas-status script from /usr/sbin and beyond that, there's a sysv initscript that periodically checks and sends an e-mail. Simple enough.

Craig

John R Pierce

5 Jan 5 Jan

3:08 a.m.

On 12/22/11 1:32 PM, Craig White wrote:

...

On Dec 22, 2011, at 1:12 PM, John R Pierce wrote:

...
...
i'm configuring a storage server with CentOS 6.2, it uses a LSI MegaRAID SAS controller, I'm using LSI's megacli to configure the storage... Any ideas on how to get drive failure notifications out of this system? I'm configuring hot spares but I'd still like some sort of notification when a drive has failed so the spare can be replaced.

don't know how to do it on CentOS but on Ubuntu, I use megaclisas-status package which goes hand in hand with megacli and it sends notifications.

If you want, I can e-mail you the megaclisas-status script from /usr/sbin and beyond that, there's a sysv initscript that periodically checks and sends an e-mail. Simple enough.

not having much luck locating that megaclisas-status script

http://hwraid.le-vert.net/wiki/LSIMegaRAIDSAS talks about it, but the source is nowhere to be found

I found this, which looks moderately interesting, but I'm not a python programmer http://windowsmasher.wordpress.com/2011/08/15/using-megacli-to-monitor-openf...

-- john r pierce N 37, W 122 santa cruz ca mid-left coast

Craig White

4:14 p.m.

On Jan 4, 2012, at 7:08 PM, John R Pierce wrote:

...

On 12/22/11 1:32 PM, Craig White wrote:

...
On Dec 22, 2011, at 1:12 PM, John R Pierce wrote:

...
...
i'm configuring a storage server with CentOS 6.2, it uses a LSI MegaRAID SAS controller, I'm using LSI's megacli to configure the storage... Any ideas on how to get drive failure notifications out of this system? I'm configuring hot spares but I'd still like some sort of notification when a drive has failed so the spare can be replaced.

don't know how to do it on CentOS but on Ubuntu, I use megaclisas-status package which goes hand in hand with megacli and it sends notifications.

If you want, I can e-mail you the megaclisas-status script from /usr/sbin and beyond that, there's a sysv initscript that periodically checks and sends an e-mail. Simple enough.

not having much luck locating that megaclisas-status script

http://hwraid.le-vert.net/wiki/LSIMegaRAIDSAS talks about it, but the source is nowhere to be found

---- It seems to me that the megaclisas-statusd is basically 2 script files, 1 in /usr/sbin and the other a sysv initscript and should be easily modifiable to run on CentOS instead of Debian/Ubuntu with the requirement that you would need to have megacli installed/running. I agree that I couldn't find any rpm's of the same.

Craig

John R Pierce

8:53 p.m.

On 01/05/12 7:14 AM, Craig White wrote:

...

...
not having much luck locating that megaclisas-status script

...
http://hwraid.le-vert.net/wiki/LSIMegaRAIDSAS talks about it, but the source is nowhere to be found

It seems to me that the megaclisas-statusd is basically 2 script files, 1 in /usr/sbin and the other a sysv initscript and should be easily modifiable to run on CentOS instead of Debian/Ubuntu with the requirement that you would need to have megacli installed/running. I agree that I couldn't find any rpm's of the same.

indeed it does. but I can't find those two scripts at the above site or elsewhere. I don't have any debian, and don't really want to have to install it.

I spent about 2 hours with the code from http://windowsmasher.wordpress.com/2011/08/15/using-megacli-to-monitor-openf... but its badly broken, the blog seems to have trashed pythons indentation, and while I fixed enough to get it working, its not working right as its not listing any of the LD or PD's.

-- john r pierce N 37, W 122 santa cruz ca mid-left coast

Craig White

9:42 p.m.

sent to you via PM - hope you don't mind.

Craig

On Jan 5, 2012, at 12:53 PM, John R Pierce wrote:

...

On 01/05/12 7:14 AM, Craig White wrote:

...
...
not having much luck locating that megaclisas-status script

...
http://hwraid.le-vert.net/wiki/LSIMegaRAIDSAS talks about it, but the source is nowhere to be found

It seems to me that the megaclisas-statusd is basically 2 script files, 1 in /usr/sbin and the other a sysv initscript and should be easily modifiable to run on CentOS instead of Debian/Ubuntu with the requirement that you would need to have megacli installed/running. I agree that I couldn't find any rpm's of the same.

indeed it does. but I can't find those two scripts at the above site or elsewhere. I don't have any debian, and don't really want to have to install it.

I spent about 2 hours with the code from http://windowsmasher.wordpress.com/2011/08/15/using-megacli-to-monitor-openf... but its badly broken, the blog seems to have trashed pythons indentation, and while I fixed enough to get it working, its not working right as its not listing any of the LD or PD's.

John R Pierce

10:02 p.m.

On 01/05/12 12:42 PM, Craig White wrote:

...

sent to you via PM - hope you don't mind.

thanks, got it.

-- john r pierce N 37, W 122 santa cruz ca mid-left coast

Craig White

10:09 p.m.

On Jan 5, 2012, at 2:02 PM, John R Pierce wrote:

...

On 01/05/12 12:42 PM, Craig White wrote:

...
sent to you via PM - hope you don't mind.

thanks, got it.

---- I probably should have figured out a way to send you just a tarball but there really wasn't much else and certainly nothing of significance. This was the full manifest...

# dpkg -L megaclisas-status /. /etc /etc/init.d /etc/init.d/megaclisas-statusd /usr /usr/sbin /usr/sbin/megaclisas-status /usr/share /usr/share/doc /usr/share/doc/megaclisas-status /usr/share/doc/megaclisas-status/README.Debian /usr/share/doc/megaclisas-status/changelog.gz /usr/share/doc/megaclisas-status/copyright

If there's anything else you want from this, let me know.

Craig

John R Pierce

10:35 p.m.

On 01/05/12 1:09 PM, Craig White wrote:

...

I probably should have figured out a way to send you just a tarball but there really wasn't much else and certainly nothing of significance.

sigh, its doing the same thing as the code I 'fixed' from that blog I posted earlier...

# ./megaclisas-status -- Controller informations -- -- ID | Model c0 | LSI MegaRAID SAS 9261-8i

-- Arrays informations -- -- ID | Type | Size | Status | InProgress

-- Disks informations -- ID | Model | Status

(I have 1 array and 36 disks on this controller) so the output format from MegaCli has probably changed just enough to throw it off, so I need to refactor it. meh.

I can't believe noone is running a late model MegaRAID SAS card with Linux and doesn't require error status change notifications.

-- john r pierce N 37, W 122 santa cruz ca mid-left coast

Craig White

6 Jan 6 Jan

12:23 a.m.

On Jan 5, 2012, at 2:35 PM, John R Pierce wrote:

...

On 01/05/12 1:09 PM, Craig White wrote:

...
I probably should have figured out a way to send you just a tarball but there really wasn't much else and certainly nothing of significance.

sigh, its doing the same thing as the code I 'fixed' from that blog I posted earlier...

# ./megaclisas-status -- Controller informations -- -- ID | Model c0 | LSI MegaRAID SAS 9261-8i

-- Arrays informations -- -- ID | Type | Size | Status | InProgress

-- Disks informations -- ID | Model | Status

(I have 1 array and 36 disks on this controller) so the output format from MegaCli has probably changed just enough to throw it off, so I need to refactor it. meh.

I can't believe noone is running a late model MegaRAID SAS card with Linux and doesn't require error status change notifications.

---- maybe it's the RAID controller you are using that isn't compatible with megacli or the version of megacli that you are using...

from dpkg -l megacli... megacli 5.00.12-1 LSI Logic MegaRAID SAS MegaCLI

/usr/sbin/megaclisas-status -- Controller informations -- -- ID | Model c0 | Supermicro SMC2108

of course this is on Ubuntu so YMMV

Craig

John R Pierce

12:52 a.m.

On 01/05/12 3:23 PM, Craig White wrote:

...

maybe it's the RAID controller you are using that isn't compatible with megacli or the version of megacli that you are using...

brand new 9261-8i SAS2 controller definitely works 100% with the latest MegaCLI64 I got from LSI Logic (installed from an RPM).... from what I'm gathering, megacli the program simply passes the command line to the card's firmware, which does all the processing and output generation, and this output format is human readable, resulting in parsing nightmares when it changes.

# rpm -qf /opt/MegaRAID/MegaCli/MegaCli64 MegaCli-8.02.16-1.i386

# /opt/MegaRAID/MegaCli/MegaCli64 showsummary a0

System Operating System: Linux version 2.6.32-220.el6.x86_64 Driver Version: 00.00.05.40-rh2 CLI Version: 8.02.16

Hardware Controller ProductName : LSI MegaRAID SAS 9261-8i(Bus 0, Dev 0) SAS Address : 500605b0032943d0 FW Package Version: 12.12.0-0046 Status : Optimal (300 more lines of drives and stuff deleted)

anyways, I think I should take this discussion off this CentOS mail list, as its really not CentOS specific, its an LSI Logic generic problem. I've got enough info now to figure out how to do my own parser, I believe I'll use that ShowSummary command rather than the LDInfo/PDInfo commands used by the other scripts.

-- john r pierce N 37, W 122 santa cruz ca mid-left coast

John R Pierce

12:34 a.m.

On 01/05/12 1:35 PM, John R Pierce wrote:

...

sigh, its doing the same thing as the code I 'fixed' from that blog I posted earlier...

ok, I've figured out the differences between what megaclisas-status expected and what megacli for these new sas cards generated, and hacked up the code to work with the version firmware I have, but it didn't really understand SAS enclosure numbering nor did it list the global hotspares:

# ./megaclisas-status -- Controller informations -- -- ID | Model c0 | LSI MegaRAID SAS 9261-8i

(note all the disks are c0u0, which is way wrong, my disks are 20:0-20:11 and 45:0-45:23)

so I took the OTHER one I'd found and saw that it WAS setup for SAS enclosure info, and on my test system generates this output...

# ./lsi-raidinfo -- Controllers -- -- ID | Model c0 | LSI MegaRAID SAS 9261-8i

20: or 45: is the enclosure/backplane and :nn is the drive on the backplane.

MEH, thats still messed up. the first hotspare above should be 20:11, and looking at the incredibly verbose output of the megacli command that this util is parsing, its a little messed up, and I think its getting the last enclosure# from a non-hotspare as they were listed first. $#@$@#$#$@#$

GAHHH! this is a mess.

-- john r pierce N 37, W 122 santa cruz ca mid-left coast

Joseph L. Casale

23 Dec 23 Dec

12:28 a.m.

...

i'm configuring a storage server with CentOS 6.2, it uses a LSI MegaRAID SAS controller, I'm using LSI's megacli to configure the storage... Any ideas on how to get drive failure notifications out of this system? I'm configuring hot spares but I'd still like some sort of notification when a drive has failed so the spare can be replaced.

I've tried countless times to get the LSI snmp agent designed just for this to work in CentOS 5 and 6, no luck.

Resorted to a pathetic snmp extend that runs megacli, sigh...

John R Pierce

12:56 a.m.

On 12/22/11 3:28 PM, Joseph L. Casale wrote:

...

...
i'm configuring a storage server with CentOS 6.2, it uses a LSI MegaRAID

...
SAS controller, I'm using LSI's megacli to configure the storage... Any ideas on how to get drive failure notifications out of this system? I'm configuring hot spares but I'd still like some sort of notification when a drive has failed so the spare can be replaced.

I've tried countless times to get the LSI snmp agent designed just for this to work in CentOS 5 and 6, no luck.

Resorted to a pathetic snmp extend that runs megacli, sigh...

I just installed the Linux MSM package from LSI...

I had to add these packages to a minimal 6.2 install for what its worth...

yum install libstdc++.i686 compat-libstdc++-33.i686 libX11 libX11.i686 libXi libXi.i686 libXtst libXtst.i686 xauth net-snmp net-snmp-utils csh

The X stuff was required by the GUI manager, I suspect if you did a agent-only install, that wouldn't be needed.

I'm a total novice when it comes to snmp. but i'm reading some howto's and stumbling around.

-- john r pierce N 37, W 122 santa cruz ca mid-left coast

4726

Age (days ago)

4740

Last active (days ago)

discuss@lists.centos.org

13 comments

3 participants

tags (0)

participants (3)

Craig White
John R Pierce
Joseph L. Casale