Okie.. that directory was, in fact, mounted RW in the RO path. Interesting!
Blind sided me a bit there :)

Let me ask, is this what happens if you use the -rw flag to fs mkmount? I
just re read the man page and.. it does seem to be true.


On Fri, Dec 20, 2013 at 9:10 PM, Timothy Balcer <[email protected]> wrote:

> server 1.6.1 and client 1.6.2, client 1.6.1 both ubuntu and CentOS
> standard distribution. It was tested from multiple clients.
>
> VLServer is 1.6.1 ubuntu.
>
> As I said previously, it was evaluated from the RO path.
>
> /afs/domain.com/photos
>
> to be specific. No 'dot'.
>
> Thanks in advance for any help :-) I am looking for troubleshooting steps
> to go beyond this.
>
> Hmm... I'll do a bit more looking. I didn't realize that any RW volume in
> the path would cause all lookups to be RW. However, as far as I know, I
> haven't mounted anything in that path RW on the RO path.
>
>
>
> On Fri, Dec 20, 2013 at 7:42 PM, Jeffrey Altman <
> [email protected]> wrote:
>
>> Basic questions.
>>
>> Which client platform and version?
>>
>> Is the mount point to the volume being evaluated from a rw volume
>> instance?  Once a rw volume is encountered in the path all volume access
>> will be to the rw volume.  (unless a mount point is explicitly to the
>> volume.readonly.)
>>
>>
>>
>>
>> On 12/20/2013 7:11 PM, Timothy Balcer wrote:
>> > ˇTo be more clear.. the problematic server (one I was salvaging) had thr
>> > RW/RO pair, and was pref 5005. I set it to 35000 through fs
>> > getserverprefs > /tmp/foo, editing that file, and reissuing it through
>> > fs setserverprefs -stdin. The other Fileservers which had RO copies were
>> > now at about 20000.
>> >
>> > It still went for the 35000 pref server.
>> >
>> >
>> > On Fri, Dec 20, 2013 at 3:10 PM, Timothy Balcer <[email protected]
>> > <mailto:[email protected]>> wrote:
>> >
>> >     I took down a file server for salvaging today, part of routine
>> >     maintenance. It had a RW/RO pair on it of volume 'photos', and there
>> >     were two RO copies on two other fileservers (on other
>> >     networks).These RO copies are mounted and online.
>> >
>> >     When the salvage process started, nothing could talk to the RO path
>> >     to the mount. Got some 'Device doesn't exist' sort of errors. Even
>> >     on the remote network fileservers.
>> >
>> >     VLDB entries are clean. Releases are all congruent.
>> >
>> >     Tried manually setting pref via fs setserverprefs.. set that server
>> >     pref to 35000
>> >
>> >     Still tries to access the local RW/RO volume for the RO path.. doing
>> >     an strace showed this was true.
>> >
>> >     The problem will fix itself once the salvage is over, but this is
>> >     disturbing... what can I do to avoid this sort of problem in future?
>> >     Am I doing something wrong here? Better troubleshooting? Advice
>> >     appreciated :)
>> >
>> >     --
>> >     Timothy Balcer / IT Services
>> >     Telmate / San Francisco, CA
>> >     Direct / (415) 300-4313 <tel:%28415%29%20300-4313>
>> >     Customer Service / (800) 205-5510
>> >
>> >
>> >
>> >
>> > --
>> > Timothy Balcer / IT Services
>> > Telmate / San Francisco, CA
>> > Direct / (415) 300-4313
>> > Customer Service / (800) 205-5510
>>
>>
>
>
> --
> Timothy Balcer / IT Services
> Telmate / San Francisco, CA
> Direct / (415) 300-4313
> Customer Service / (800) 205-5510
>



-- 
Timothy Balcer / IT Services
Telmate / San Francisco, CA
Direct / (415) 300-4313
Customer Service / (800) 205-5510

Reply via email to