let's keep the info on the list, more peple more ideas

-------- Original Message --------
Subject:        Re: [Freeipa-users] IPA RUV unable to decode
Date:   Tue, 5 May 2015 18:32:15 +0200
From:   Vaclav Adamec <vaclav.ada...@suchy-zleb.cz>
To:     Ludwig Krispenz <lkris...@redhat.com>

ipa-replica-manage del <replica_server> -fc
ipa-replica-prepare ... + copy gpg

ipa-server-install --uninstall
ipa-replica-install ...

one by one on all replicas (1 CA master and 5 replicas which are also set in DNS as SRV records). Plus whole the time there was script/job which register clients (ipa-client-install --uninstall + ipa-client-install ... --force-join .... different active replica servers - via DNS). Seems that I have some issue with replica itself, it seems to work, but when it's loaded with such "heavy" operation replica servers goes down (dir process died) even master itself went down with minor codes errs in log.

On Tue, May 5, 2015 at 3:25 PM, Ludwig Krispenz <lkris...@redhat.com <mailto:lkris...@redhat.com>> wrote:

   On 05/05/2015 02:24 PM, Vaclav Adamec wrote:
    For me it was quite easy to reached this state, I just setup
    replica on 6 servers (two datacenters, with ~15ms latency, same
    stable ntp source), run test scripts which add/verify random
    account, add/remove hosts (unregister seems to be "heaviest"
    operation to replicas). During these tests I removed some remove
    servers from replica and try to add them back,
   what exactly did you do:
   ipa-server-install --uninstall
   ipa-server-install ?
   any other commands ?

    result is like this. Seems that with higher load it happens more
    often (register/unregister multiple servers, with actual setup it
    is about 10 new registration in same time to different replica
    servers), also sometimes I got replication conflicts or dir
    service complete failure (dirsrv down, started with gssapi minor
    code error). But even If I stop tests queue is not freed, RUV
    still there (old replicas).

    On Tue, May 5, 2015 at 1:49 PM, Ludwig Krispenz
    <lkris...@redhat.com <mailto:lkris...@redhat.com>> wrote:

        On 05/05/2015 01:27 PM, Martin Kosek wrote:

            On 05/05/2015 12:38 PM, Vaclav Adamec wrote:

                  I tried migrate to newest version IPA, but result is
                quite unstable and
                removing old replicas ends with RUV which cannot be
                decoded (it stucked in
                queue forever):

                ipa-replica-manage del ipa-master-dmz002.test.com
                <http://ipa-master-dmz002.test.com> -fc
                Cleaning a master is irreversible.
                This should not normally be require, so use cautiously.
                Continue to clean master? [no]: yes

                ipa-replica-manage list-ruv
                unable to decode: {replica 8} 55091239000400080000
                unable to decode: {replica 7} 552f84cd000300070000
                unable to decode: {replica 11} 551a42f70000000b0000
                unable to decode: {replica 15} 551e82e10001000f0000
                unable to decode: {replica 14} 551e82ec0001000e0000
                unable to decode: {replica 20} 552f4b72000600140000
                unable to decode: {replica 10} 551a25af0001000a0000
                unable to decode: {replica 3} 551e864c000300030000
                unable to decode: {replica 5} 55083ad2000300050000
                unable to decode: {replica 9} 550913e7000000090000
                unable to decode: {replica 19} 55210193000300130000
                unable to decode: {replica 12} 551a48290000000c0000
                <http://ipa-master-dmz001.test.com:389>: 25
                <http://ipa-master-dmz002.test.com:389>: 21

                it is possible to clear this queue and leave only
                valid servers ?

                Thanks in advance


            Ludwig or Thierry, do you know? The questions about RUV
            cleaning seems to be
            recurring, I suspect there will be a pattern (bug) and not
            just configuration

        we have seen this in a recent thread, and it is clear that the
        RUV is corrupted and cannot be decoded, but we don't have a
        scenario how this is state is reached.

-- -- May the fox be with you ...
       ) )         /\_/\
      (_=---_(@ @)
        (          \   /
        /|/----\|\  V
        " "     " "

-- May the fox be with you ...
   ) )         /\_/\
  (_=---_(@ @)
    (          \   /
    /|/----\|\  V
    " "     " "

Manage your subscription for the Freeipa-users mailing list:
Go to http://freeipa.org for more info on the project

Reply via email to