fma plugin documentation

[James C. McPherson]

On Mon, 24 Aug 2009 10:00:07 +0200

Hi list,
I would like to get better a understanding for the fma plugins and its
config-files like
/usr/lib/fm/fmd/plugins/disk-transport.conf
May someone provide me a link to a detailed dokumentation? I searched
through sun.com and opensolaris.org without success.
Very frustrating...

Apart from reading the source code

http://src.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/cmd/fm/

you should have a look at the copious doco available at

http://www.opensolaris.org/os/community/fm/

It's also a very good idea to look at the hitstory log fort the
code itself, since that contains links to PSARC materials which
are the authoritative source for how things work.
Eg,
http://src.opensolaris.org/source/history/onnv/onnv-gate/usr/src/cmd/fm/dicts/DISK.dict
links to

PSARC 2008/558 SCSI Disk Device-as-Detector Diagnosis (phase3)
http://arc.opensolaris.org/caselog/PSARC/2008/558


James C. McPherson
--
Senior Kernel Software Engineer, Solaris
Sun Microsystems
http://blogs.sun.com/jmcp http://www.jmcp.homeunix.com/blog

[Gavin Maltby]

Hi,

Hi list,
I would like to get better a understanding for the fma plugins and its
config-files like
/usr/lib/fm/fmd/plugins/disk-transport.conf
May someone provide me a link to a detailed dokumentation? I searched
through sun.com and opensolaris.org without success.
Very frustrating...

You'll find the fmd PRM on the fm community website James referenced;
the interface table there will also tell you that none of the interfaces
are Committed in the formal sense - they're still evolving.
You may also find that the PRM is somewhat out-of-date
in terms of some of the more recently introduced plugins and
other functionality. If you can share what it is you want
to configure or achieve we can supply more detail if necessary.

Cheers

Gavin

[Tobias Exner]

Hi,

As I can see there seems to be no pdf-file with a simple overview with
the funtionality of the plugins and their configurable parameters.
Am I right?


I want to know 2 things


1.
Which plugins are checking physical disks and how can I stop or delay it.


2.
Using the powerd of Solaris I'm able to standby a disk after some idle time.
Reading from that disk took roundabout 14 or 24 seconds until the first
ios are accepted.
The support guys from Hitachi tells me that a disk needs 14-16 seconds
to spin up. So why tooks it longer sometimes?
Thats not critical but interesting for me.

Is there a timeout somewhere in the fma. Could the fma be responsible
for that behaviour, at all?




regards,

Tobias

Post by Gavin Maltby
Hi,

Hi list,
I would like to get better a understanding for the fma plugins and
its config-files like
/usr/lib/fm/fmd/plugins/disk-transport.conf
May someone provide me a link to a detailed dokumentation? I searched
through sun.com and opensolaris.org without success.
Very frustrating...

You'll find the fmd PRM on the fm community website James referenced;
the interface table there will also tell you that none of the interfaces
are Committed in the formal sense - they're still evolving.
You may also find that the PRM is somewhat out-of-date
in terms of some of the more recently introduced plugins and
other functionality. If you can share what it is you want
to configure or achieve we can supply more detail if necessary.
Cheers
Gavin

[Tarik Soydan]

Post by Tobias Exner
Hi,
As I can see there seems to be no pdf-file with a simple overview with
the funtionality of the plugins and their configurable parameters.
Am I right?
I want to know 2 things
1.
Which plugins are checking physical disks and how can I stop or delay it.

The "disk-transport" module will poll disks for errors periodically (1
hour default I believe).
However, unless you have an X4200, X4500, X4540, X6000 or an external SES
controlled JBOD this will not be the case since there is no disk
topology otherwise.

You should be able to use "fmadm" to unload the module if you need to.

Post by Tobias Exner
2.
Using the powerd of Solaris I'm able to standby a disk after some idle time.
Reading from that disk took roundabout 14 or 24 seconds until the
first ios are accepted.
The support guys from Hitachi tells me that a disk needs 14-16 seconds
to spin up. So why tooks it longer sometimes?
Thats not critical but interesting for me.
Is there a timeout somewhere in the fma. Could the fma be responsible
for that behaviour, at all?

I doubt it.
If anything I would think that FMA would make the disks spin up from
standby state sooner due to the periodic polling above, not delay it.

-tarik

Post by Tobias Exner
regards,
Tobias

Post by Gavin Maltby
Hi,

Hi list,
I would like to get better a understanding for the fma plugins and
its config-files like
/usr/lib/fm/fmd/plugins/disk-transport.conf
May someone provide me a link to a detailed dokumentation? I
searched through sun.com and opensolaris.org without success.
Very frustrating...

You'll find the fmd PRM on the fm community website James referenced;
the interface table there will also tell you that none of the interfaces
are Committed in the formal sense - they're still evolving.
You may also find that the PRM is somewhat out-of-date
in terms of some of the more recently introduced plugins and
other functionality. If you can share what it is you want
to configure or achieve we can supply more detail if necessary.
Cheers
Gavin

_______________________________________________
fm-discuss mailing list

[Scott Davenport]

Post by Tarik Soydan

Post by Tobias Exner
Hi,
As I can see there seems to be no pdf-file with a simple overview with
the funtionality of the plugins and their configurable parameters.
Am I right?
I want to know 2 things
1.
Which plugins are checking physical disks and how can I stop or delay it.

The "disk-transport" module will poll disks for errors periodically (1
hour default I believe).
However, unless you have an X4200, X4500, X4540, X6000 or an external SES
controlled JBOD this will not be the case since there is no disk
topology otherwise.
You should be able to use "fmadm" to unload the module if you need to.

Post by Tobias Exner
2.
Using the powerd of Solaris I'm able to standby a disk after some idle time.
Reading from that disk took roundabout 14 or 24 seconds until the
first ios are accepted.
The support guys from Hitachi tells me that a disk needs 14-16 seconds
to spin up. So why tooks it longer sometimes?
Thats not critical but interesting for me.
Is there a timeout somewhere in the fma. Could the fma be responsible
for that behaviour, at all?

I doubt it.
If anything I would think that FMA would make the disks spin up from
standby state sooner due to the periodic polling above, not delay it.
From a power management perspective, should FMA *not* poll on disks

that have been put into a standby mode?

-scott

Post by Tarik Soydan

Post by Tobias Exner
regards,
Tobias

Post by Gavin Maltby
Hi,

Hi list,
I would like to get better a understanding for the fma plugins and
its config-files like
/usr/lib/fm/fmd/plugins/disk-transport.conf
May someone provide me a link to a detailed dokumentation? I
searched through sun.com and opensolaris.org without success.
Very frustrating...

You'll find the fmd PRM on the fm community website James referenced;
the interface table there will also tell you that none of the interfaces
are Committed in the formal sense - they're still evolving.
You may also find that the PRM is somewhat out-of-date
in terms of some of the more recently introduced plugins and
other functionality. If you can share what it is you want
to configure or achieve we can supply more detail if necessary.
Cheers
Gavin

_______________________________________________
fm-discuss mailing list

_______________________________________________
fm-discuss mailing list

[Tarik Soydan]

Post by Scott Davenport

Post by Tarik Soydan

Post by Tobias Exner
Hi,
As I can see there seems to be no pdf-file with a simple overview with
the funtionality of the plugins and their configurable parameters.
Am I right?
I want to know 2 things
1.
Which plugins are checking physical disks and how can I stop or delay it.

The "disk-transport" module will poll disks for errors periodically (1
hour default I believe).
However, unless you have an X4200, X4500, X4540, X6000 or an external SES
controlled JBOD this will not be the case since there is no disk
topology otherwise.
You should be able to use "fmadm" to unload the module if you need to.

Post by Tobias Exner
2.
Using the powerd of Solaris I'm able to standby a disk after some idle time.
Reading from that disk took roundabout 14 or 24 seconds until the
first ios are accepted.
The support guys from Hitachi tells me that a disk needs 14-16 seconds
to spin up. So why tooks it longer sometimes?
Thats not critical but interesting for me.
Is there a timeout somewhere in the fma. Could the fma be responsible
for that behaviour, at all?

I doubt it.
If anything I would think that FMA would make the disks spin up from
standby state sooner due to the periodic polling above, not delay it.
From a power management perspective, should FMA *not* poll on disks

that have been put into a standby mode?

I would think not, but maybe we could check on SMART status
without actually causing the disk to spin up?

-tarik

Post by Scott Davenport
-scott

Post by Tarik Soydan

Post by Tobias Exner
regards,
Tobias

Post by Gavin Maltby
Hi,

Hi list,
I would like to get better a understanding for the fma plugins and
its config-files like
/usr/lib/fm/fmd/plugins/disk-transport.conf
May someone provide me a link to a detailed dokumentation? I
searched through sun.com and opensolaris.org without success.
Very frustrating...

You'll find the fmd PRM on the fm community website James referenced;
the interface table there will also tell you that none of the interfaces
are Committed in the formal sense - they're still evolving.
You may also find that the PRM is somewhat out-of-date
in terms of some of the more recently introduced plugins and
other functionality. If you can share what it is you want
to configure or achieve we can supply more detail if necessary.
Cheers
Gavin

_______________________________________________
fm-discuss mailing list

_______________________________________________
fm-discuss mailing list

[Steve Hanson]

Post by Tarik Soydan

Post by Scott Davenport

Post by Tarik Soydan

Post by Tobias Exner
Hi,
As I can see there seems to be no pdf-file with a simple overview with
the funtionality of the plugins and their configurable parameters.
Am I right?
I want to know 2 things
1.
Which plugins are checking physical disks and how can I stop or delay it.

The "disk-transport" module will poll disks for errors periodically (1
hour default I believe).
However, unless you have an X4200, X4500, X4540, X6000 or an external SES
controlled JBOD this will not be the case since there is no disk
topology otherwise.
You should be able to use "fmadm" to unload the module if you need to.

Post by Tobias Exner
2.
Using the powerd of Solaris I'm able to standby a disk after some idle time.
Reading from that disk took roundabout 14 or 24 seconds until the
first ios are accepted.
The support guys from Hitachi tells me that a disk needs 14-16 seconds
to spin up. So why tooks it longer sometimes?
Thats not critical but interesting for me.
Is there a timeout somewhere in the fma. Could the fma be responsible
for that behaviour, at all?

I doubt it.
If anything I would think that FMA would make the disks spin up from
standby state sooner due to the periodic polling above, not delay it.
From a power management perspective, should FMA *not* poll on disks

that have been put into a standby mode?

I would think not, but maybe we could check on SMART status
without actually causing the disk to spin up?

I wouldn't have thought the MODE_SENSE to get the SMART data would itself
need to spin up the disks. However it's possible that libdiskstatus might
spin the disk up when opening the device. Might be worth checking with Eric?

Steve

Post by Tarik Soydan
-tarik

Post by Scott Davenport
-scott

Post by Tarik Soydan

Post by Tobias Exner
regards,
Tobias

Post by Gavin Maltby
Hi,

Hi list,
I would like to get better a understanding for the fma plugins and
its config-files like
/usr/lib/fm/fmd/plugins/disk-transport.conf
May someone provide me a link to a detailed dokumentation? I
searched through sun.com and opensolaris.org without success.
Very frustrating...

You'll find the fmd PRM on the fm community website James referenced;
the interface table there will also tell you that none of the interfaces
are Committed in the formal sense - they're still evolving.
You may also find that the PRM is somewhat out-of-date
in terms of some of the more recently introduced plugins and
other functionality. If you can share what it is you want
to configure or achieve we can supply more detail if necessary.
Cheers
Gavin

_______________________________________________
fm-discuss mailing list

_______________________________________________
fm-discuss mailing list

------------------------------------------------------------------------
_______________________________________________
fm-discuss mailing list

[Eric Schrock]

Post by Steve Hanson

Post by Tarik Soydan

Post by Scott Davenport

Post by Tarik Soydan
If anything I would think that FMA would make the disks spin up from
standby state sooner due to the periodic polling above, not delay it.
From a power management perspective, should FMA *not* poll on disks

that have been put into a standby mode?

I would think not, but maybe we could check on SMART status
without actually causing the disk to spin up?

I wouldn't have thought the MODE_SENSE to get the SMART data would itself
need to spin up the disks. However it's possible that libdiskstatus might
spin the disk up when opening the device. Might be worth checking with Eric?

Reading the log pages will definitely require the disks to spin up.
This is a known issue with the disk-transport module (and it's
previous incarnation, sfx4500-disk). The long term plan is to replace
it entirely with in-band monitoring via CHECK CONDITION codes instead
of the polling mechanism. In the short term, it would not be
difficult to check the devinfo power management properties, expose
that through libtopo, and skip those disks that are powered down.

- Eric

--
Eric Schrock, Fishworks http://blogs.sun.com/eschrock