Bug 693948

Ben:

I need your confirmation on whether this information is actually correct. This may be a confusion over the issue of default values vs. compiled-in values.

Points 1 and 2 refer to table 4.1, path_checker and path_grouping_policy descriptions:

http://documentation-stage.bne.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/96/html/DM_Multipath/config_file_defaults.html

Point 3 refers to the multipath command output described in section 5.5:

http://documentation-stage.bne.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/96/html/DM_Multipath/MPIO_output.html

This bug gives output to the command that does seem to have one more field than is described in the document.

-Steven

Yes. The bug is correct.

In table 4.1, the default path_checker should be 'directio', and the default path_grouping_policy should be 'failover'

In section 5.5, the path format should read
 `- host:channel:id:lun devnode major:minor dm_status_if_known path_status online_status

Also the possible path_status values of a path that is working correctly are "ready" or "ghost" (not active) and the possible path_status values of a path that is not working correctly are "faulty" and "shaky" (not failed)

"active" and "failed" are returned for the dm_status_if_know, and your description of them is correct.

The possible values for online_status are "running" and "offline".  "offline" means that this scsi device has been disabled.

Thanks, Ben. And thanks Gris for finding this. I'll be able to get those corrections into the document in time for the RHEL 6.1 release.

Ben:

(Sorry to put this back in NEEDINFO, but this yields another question...)

The bug refers to table 4.1 (multipath configuration defaults), but Table 4.3 (device attributes) also gives readsector0 as the default value for path_checker. Should that be directio as well?

Table 4-2 (multipath attributes) and Table 4.3 (device attributes) give no default value for path_grouping_policy. Should I note a default value of failover in those tables?

Table 4-2:
http://documentation-stage.bne.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/96/html/DM_Multipath/config_file_multipath.html

Table 4-3:
http://documentation-stage.bne.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/96/html/DM_Multipath/config_file_devices.html

-Steven

I think I forgot to specify Ben for the NEEDINFO...

Actually, I don't think I would include defaults in the devices or multipaths attribute tables at all.  The way multipath.conf works, if a device has a multipaths section, and the attribute is specified there, that value is used. Otherwise, if the device has a devices section and the attribute is specified there, that value is used.  Otherwise if there is a defaults section and the attribute is specified there, that one is used. If the value is not configured anywhere, then the default is used.

Putting defaults in the devices and multipaths sections seems a little misleading, because if a user doesn't set an attribute there, mutipath just checks for it at a more general level.  If it is set in the more general level, then multipath won't be using the default.

In my current working 6.1 draft of the DM-Multipath manual, I have corrected the errors noted in this bug and confirmed by Ben. I have also removed the default values from tables 4-2 and 4-3. This updates will be reflected in the RHEL 6.1 release.

The manual with the corrections has been checked into SVN and built on the Brew server so I am moving this bug to MODIFIED.

Red_Hat_Enterprise_Linux-DM_Multipath-6-web-en-US-1-38.el5

The default values for path_checker and path_grouping_policy have been fixed in Table 4-1:

http://documentation-stage.bne.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/6/html/DM_Multipath/config_file_defaults.html

Section 5.7 now includes the corrected path format output:

http://documentation-stage.bne.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/6/html/DM_Multipath/MPIO_output.html

Verified: Red_Hat_Enterprise_Linux-DM_Multipath-6-web-en-US-1-38.el5

(In reply to comment #2)
> Yes. The bug is correct.
> 
> In table 4.1, the default path_checker should be 'directio', and the default
> path_grouping_policy should be 'failover'
> 

fixed

> In section 5.5, the path format should read
>  `- host:channel:id:lun devnode major:minor dm_status_if_known path_status
> online_status
> 

fixed

> Also the possible path_status values of a path that is working correctly are
> "ready" or "ghost" (not active) and the possible path_status values of a path
> that is not working correctly are "faulty" and "shaky" (not failed)
> 

fixed: "the status of the path is ready or ghost. If the path is down, the status is faulty or shaky." 

> "active" and "failed" are returned for the dm_status_if_know, and your
> description of them is correct.
> 
> The possible values for online_status are "running" and "offline".  "offline"
> means that this scsi device has been disabled.

fixed: "The possible values for online_status are running and offline. A status of offline means that this scsi device has been disabled. "

With the release of RHEL 6.1, I am closing this bug.