6:43 a.m.
Hello everyone,
I have been testing n-way multimaster replication with OpenLDAP for a while
now (from 2.4.11, to 2.4.15) and just when I though that everything was
working perfectly, I dicided to test N-way multimaster not only with 2 masters
on different servers, but with 4 !
2 OpenLDAP instances per server.
I have been configuring syncprov and syncrepl accordingly :
olcServerID: 1 ldap://163.106.38.90:9011/
olcServerID: 2 ldap://163.106.38.92:9012/
olcServerID: 3 ldap://163.106.38.90:9013/
olcServerID: 4 ldap://163.106.38.92:9014/
olcSyncrepl: {0}rid=011 provider=ldap://163.106.38.90:9011/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {1}rid=012 provider=ldap://163.106.38.92:9012/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {2}rid=013 provider=ldap://163.106.38.90:9013/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {3}rid=014 provider=ldap://163.106.38.92:9014/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
I am starting with all instances synced and I am trying to add entries en all
four instances (in //). If I do so, I have a few entries that are not
replicated on the others. I am getting this kind of messages :
do_syncrep2:
cookie=rid=011,sid=002,csn=20090227130003.849482Z#000000#004#000000
do_syncrep2: rid=011 CSN too old, ignoring
20090227130003.849482Z#000000#004#000000
do_syncrep2:
cookie=rid=013,sid=002,csn=20090227130003.849482Z#000000#004#000000
do_syncrep2: rid=013 CSN too old, ignoring
20090227130003.849482Z#000000#004#000000
do_syncrep2:
cookie=rid=014,sid=002,csn=20090227130003.946474Z#000000#004#000000
Did someone face the same issue ?
Here is my configuration : (I am using refreshAndPersist mode for both cn=config
and olcDatabase={1}bdb)
M1 on IP1 / PORT1 :
dn: cn=config
objectClass: olcGlobal
cn: config
structuralObjectClass: olcGlobal
creatorsName: cn=config
olcServerID: 1 ldap://163.106.38.90:9011/
olcServerID: 2 ldap://163.106.38.92:9012/
olcServerID: 3 ldap://163.106.38.90:9013/
olcServerID: 4 ldap://163.106.38.92:9014/
entryUUID: ef89c876-adb3-4dc7-aa7d-024bbc359c98
createTimestamp: 20090227085748Z
entryCSN: 20090227085749.920499Z#000000#004#000000
modifiersName: cn=config
modifyTimestamp: 20090227085749Z
contextCSN: 20090227085752.833630Z#000000#001#000000
dn: olcDatabase={1}bdb
objectClass: olcDatabaseConfig
objectClass: olcBdbConfig
olcDatabase: {1}bdb
olcDbDirectory: ./openldap-data
olcSuffix: c=fr
olcRootDN: cn=admin,c=fr
olcRootPW:: e1NTSEF9WVZNSHJtYTRvUGd4KzFoak9kYWhBcm5NVHJxU1Zmdno=
olcSizeLimit: 100
olcSyncrepl: {0}rid=011 provider=ldap://163.106.38.90:9011/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {1}rid=012 provider=ldap://163.106.38.92:9012/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {2}rid=013 provider=ldap://163.106.38.90:9013/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {3}rid=014 provider=ldap://163.106.38.92:9014/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcTimeLimit: 600
olcMirrorMode: TRUE
olcDbCacheSize: 2000
olcDbCheckpoint: 2000 10
olcDbIndex: default pres,eq
olcDbIndex: cn,sn pres,eq,sub
olcDbIndex: objectClass,entryCSN,entryUUID eq
structuralObjectClass: olcBdbConfig
entryUUID: 00c01e5d-69ee-4baa-8e5a-4ef609dfd958
creatorsName: cn=config
createTimestamp: 20090227085752Z
entryCSN: 20090227085752.729899Z#000000#001#000000
modifiersName: cn=config
modifyTimestamp: 20090227085752Z
M2 on IP2 / PORT2 :
dn: cn=config
objectClass: olcGlobal
cn: config
structuralObjectClass: olcGlobal
entryUUID: 8da75037-65e6-4375-8c21-7e5c0194a60b
creatorsName: cn=config
createTimestamp: 20090227085723Z
olcServerID: 1 ldap://163.106.38.90:9011/
olcServerID: 2 ldap://163.106.38.92:9012/
olcServerID: 3 ldap://163.106.38.90:9013/
olcServerID: 4 ldap://163.106.38.92:9014/
entryCSN: 20090227085725.003182Z#000000#002#000000
modifiersName: cn=config
modifyTimestamp: 20090227085725Z
contextCSN: 20090227085752.833630Z#000000#001#000000
dn: olcDatabase={1}bdb
objectClass: olcDatabaseConfig
objectClass: olcBdbConfig
olcDatabase: {1}bdb
olcDbDirectory: ./openldap-data
olcSuffix: c=fr
olcRootDN: cn=admin,c=fr
olcRootPW:: e1NTSEF9WVZNSHJtYTRvUGd4KzFoak9kYWhBcm5NVHJxU1Zmdno=
olcSizeLimit: 100
olcSyncrepl: {0}rid=011 provider=ldap://163.106.38.90:9011/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {1}rid=012 provider=ldap://163.106.38.92:9012/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {2}rid=013 provider=ldap://163.106.38.90:9013/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {3}rid=014 provider=ldap://163.106.38.92:9014/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcTimeLimit: 600
olcMirrorMode: TRUE
olcDbCacheSize: 2000
olcDbCheckpoint: 2000 10
olcDbIndex: default pres,eq
olcDbIndex: cn,sn pres,eq,sub
olcDbIndex: objectClass,entryCSN,entryUUID eq
structuralObjectClass: olcBdbConfig
entryUUID: 00c01e5d-69ee-4baa-8e5a-4ef609dfd958
creatorsName: cn=config
createTimestamp: 20090227085752Z
entryCSN: 20090227085752.729899Z#000000#001#000000
modifiersName: cn=config
modifyTimestamp: 20090227085752Z
M3 on IP1 / PORT3 :
dn: cn=config
objectClass: olcGlobal
cn: config
structuralObjectClass: olcGlobal
entryUUID: cf068647-318f-4848-9c72-9c7745a8a4b3
creatorsName: cn=config
createTimestamp: 20090227085742Z
olcServerID: 1 ldap://163.106.38.90:9011/
olcServerID: 2 ldap://163.106.38.92:9012/
olcServerID: 3 ldap://163.106.38.90:9013/
olcServerID: 4 ldap://163.106.38.92:9014/
entryCSN: 20090227085743.825685Z#000000#003#000000
modifiersName: cn=config
modifyTimestamp: 20090227085743Z
contextCSN: 20090227085752.833630Z#000000#001#000000
dn: olcDatabase={1}bdb
objectClass: olcDatabaseConfig
objectClass: olcBdbConfig
olcDatabase: {1}bdb
olcDbDirectory: ./openldap-data
olcSuffix: c=fr
olcRootDN: cn=admin,c=fr
olcRootPW:: e1NTSEF9WVZNSHJtYTRvUGd4KzFoak9kYWhBcm5NVHJxU1Zmdno=
olcSizeLimit: 100
olcSyncrepl: {0}rid=011 provider=ldap://163.106.38.90:9011/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {1}rid=012 provider=ldap://163.106.38.92:9012/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {2}rid=013 provider=ldap://163.106.38.90:9013/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {3}rid=014 provider=ldap://163.106.38.92:9014/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcTimeLimit: 600
olcMirrorMode: TRUE
olcDbCacheSize: 2000
olcDbCheckpoint: 2000 10
olcDbIndex: default pres,eq
olcDbIndex: cn,sn pres,eq,sub
olcDbIndex: objectClass,entryCSN,entryUUID eq
structuralObjectClass: olcBdbConfig
entryUUID: 00c01e5d-69ee-4baa-8e5a-4ef609dfd958
creatorsName: cn=config
createTimestamp: 20090227085752Z
entryCSN: 20090227085752.729899Z#000000#001#000000
modifiersName: cn=config
modifyTimestamp: 20090227085752Z
M4 on IP2 / PORT4 :
dn: cn=config
objectClass: olcGlobal
cn: config
structuralObjectClass: olcGlobal
entryUUID: ef89c876-adb3-4dc7-aa7d-024bbc359c98
creatorsName: cn=config
createTimestamp: 20090227085748Z
olcServerID: 1 ldap://163.106.38.90:9011/
olcServerID: 2 ldap://163.106.38.92:9012/
olcServerID: 3 ldap://163.106.38.90:9013/
olcServerID: 4 ldap://163.106.38.92:9014/
entryCSN: 20090227085749.920499Z#000000#004#000000
modifiersName: cn=config
modifyTimestamp: 20090227085749Z
contextCSN: 20090227085752.833630Z#000000#001#000000
dn: olcDatabase={1}bdb
objectClass: olcDatabaseConfig
objectClass: olcBdbConfig
olcDatabase: {1}bdb
olcDbDirectory: ./openldap-data
olcSuffix: c=fr
olcRootDN: cn=admin,c=fr
olcRootPW:: e1NTSEF9WVZNSHJtYTRvUGd4KzFoak9kYWhBcm5NVHJxU1Zmdno=
olcSizeLimit: 100
olcSyncrepl: {0}rid=011 provider=ldap://163.106.38.90:9011/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {1}rid=012 provider=ldap://163.106.38.92:9012/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {2}rid=013 provider=ldap://163.106.38.90:9013/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcSyncrepl: {3}rid=014 provider=ldap://163.106.38.92:9014/ binddn="cn=admin,c
=fr" bindmethod=simple credentials=secret searchbase="c=fr"
type=refreshAndPe
rsist retry="5 5 300 12 3600 +" timeout=3
olcTimeLimit: 600
olcMirrorMode: TRUE
olcDbCacheSize: 2000
olcDbCheckpoint: 2000 10
olcDbIndex: default pres,eq
olcDbIndex: cn,sn pres,eq,sub
olcDbIndex: objectClass,entryCSN,entryUUID eq
structuralObjectClass: olcBdbConfig
entryUUID: 00c01e5d-69ee-4baa-8e5a-4ef609dfd958
creatorsName: cn=config
createTimestamp: 20090227085752Z
entryCSN: 20090227085752.729899Z#000000#001#000000
modifiersName: cn=config
modifyTimestamp: 20090227085752Z
I also have to say that if I stop M3 & M4 and try to add entries to M1 & M2, I
don't get the "CSN too old" messages and all entries a replicated correctly
!!!
Is this because Multimaster is limited to 2-3 instances ? The reason why I am
trying to build a 4-way master architecture is because I would like to be able
to stop one server and perform a slapcat, even if one "physical" server is
down...
Thanks very much for your consideration & time..
Adrien Futschik
9:37 a.m.
--On Friday, February 27, 2009 3:43 PM +0100 Adrien Futschik
<adrien.futschik(a)atosorigin.com> wrote:
Hello everyone,
I also have to say that if I stop M3 & M4 and try to add entries
to M1 &
M2, I don't get the "CSN too old" messages and all entries a replicated
correctly !!!
Is this because Multimaster is limited to 2-3 instances ? The reason why
I am trying to build a 4-way master architecture is because I would like
to be able to stop one server and perform a slapcat, even if one
"physical" server is down...
N-Way means for any value of N. So 4 should work just fine. Are all 4 of
your servers tightly synchronized, time-wise? I.e., using ntp to make sure
their clocks all agree?
--Quanah
--
Quanah Gibson-Mount
Principal Software Engineer
Zimbra, Inc
--------------------
Zimbra :: the leader in open source messaging and collaboration
5:20 a.m.
Le vendredi 27 février 2009 18:37:33, Quanah Gibson-Mount a écrit :
--On Friday, February 27, 2009 3:43 PM +0100 Adrien Futschik
<adrien.futschik(a)atosorigin.com> wrote:
> Hello everyone,
>
> I also have to say that if I stop M3 & M4 and try to add entries to M1 &
> M2, I don't get the "CSN too old" messages and all entries a
replicated
> correctly !!!
>
> Is this because Multimaster is limited to 2-3 instances ? The reason why
> I am trying to build a 4-way master architecture is because I would like
> to be able to stop one server and perform a slapcat, even if one
> "physical" server is down...
N-Way means for any value of N. So 4 should work just fine. Are all 4 of
your servers tightly synchronized, time-wise? I.e., using ntp to make sure
their clocks all agree?
--Quanah
--
Quanah Gibson-Mount
Principal Software Engineer
Zimbra, Inc
--------------------
Zimbra :: the leader in open source messaging and collaboration
Yes, they are synchronized with NTP !
5:39 a.m.
Le vendredi 27 février 2009 18:37:33, Quanah Gibson-Mount a écrit :
--On Friday, February 27, 2009 3:43 PM +0100 Adrien Futschik
<adrien.futschik(a)atosorigin.com> wrote:
> Hello everyone,
>
> I also have to say that if I stop M3 & M4 and try to add entries to M1 &
> M2, I don't get the "CSN too old" messages and all entries a
replicated
> correctly !!!
>
> Is this because Multimaster is limited to 2-3 instances ? The reason why
> I am trying to build a 4-way master architecture is because I would like
> to be able to stop one server and perform a slapcat, even if one
> "physical" server is down...
N-Way means for any value of N. So 4 should work just fine. Are all 4 of
your servers tightly synchronized, time-wise? I.e., using ntp to make sure
their clocks all agree?
--Quanah
--
Quanah Gibson-Mount
Principal Software Engineer
Zimbra, Inc
--------------------
Zimbra :: the leader in open source messaging and collaboration
Considering that M1 & M3 are on the same server and therefore have exactly the
same time, if this was a time related problem, I should'nt get any "CSN too
old" messages between M1&M3 and M2&M4, should I ?
I have also noticed that when M1 gets a new entry and passes it to M2&M3&M4,
when M2&M3&M4 revieve it, they also pass it to M2&M3&M4 ! I don't
understand
why this happends but it look's very much like this is what's happening,
because sometimes, M2 would have passed-it to M4, before M4 has actualy
revieved the add order from M1.
I therefore happend to notice that sometimes, entries send from M1 are
revieved in the wrong ordrer by other masters and therefore some entries may
be skipped !!!
Here is a example :
I add cn=M1client1 & cn=M1client2 on M1,
M1client1 & M1client2 are successuly replicated on M2&M4 but on M3, only
M1client2 is inserted and I am getting an "CSN too old" message for M1client1
on M3.
I don't have the logfile here, I'll send extracts this monday.
I am also getting this messages from time to time :
=> bdb_idl_insert_key: c_put id failed: DB_LOCK_DEADLOCK: Locker killed to
resolve a deadlock (-30994)
=> bdb_dn2id_add 0x1e40: parent (ou=clients,o=edf,c=fr) insert failed: -30994
I guess this is because all 4 masters recieve entries that have the same
parent : ou=clients,o=edf,c=fr and that happends if two entries are "inserted"
simultaniously.
Regards
Adrien Futschik
3:18 a.m.
Adrien Futschik wrote:
Considering that M1& M3 are on the same server and therefore
have exactly the
same time, if this was a time related problem, I should'nt get any "CSN too
old" messages between M1&M3 and M2&M4, should I ?
I have also noticed that when M1 gets a new entry and passes it to M2&M3&M4,
when M2&M3&M4 revieve it, they also pass it to M2&M3&M4 ! I don't
understand
why this happends but it look's very much like this is what's happening,
because sometimes, M2 would have passed-it to M4, before M4 has actualy
revieved the add order from M1.
I therefore happend to notice that sometimes, entries send from M1 are
revieved in the wrong ordrer by other masters and therefore some entries may
be skipped !!!
Yes, that makes sense. The CSN check assumes changes will always be received
in the same order they were sent from the provider. Obviously in this case
this assumption is wrong. You should submit an ITS for this.
This problem was discussed on the -devel list back in 2007; the code ought to
be using a spanning tree/routing algorithm to ensure that when multiple routes
exist for propagating a change, the change is delivered exactly once.
Unfortunately no one has spent any further time on this issue since then.
Here is a example :
I add cn=M1client1& cn=M1client2 on M1,
M1client1& M1client2 are successuly replicated on M2&M4 but on M3, only
M1client2 is inserted and I am getting an "CSN too old" message for M1client1
on M3.
I don't have the logfile here, I'll send extracts this monday.
I am also getting this messages from time to time :
=> bdb_idl_insert_key: c_put id failed: DB_LOCK_DEADLOCK: Locker killed to
resolve a deadlock (-30994)
=> bdb_dn2id_add 0x1e40: parent (ou=clients,o=edf,c=fr) insert failed: -30994
I guess this is because all 4 masters recieve entries that have the same
parent : ou=clients,o=edf,c=fr and that happends if two entries are "inserted"
simultaniously.
DB_LOCK_DEADLOCK messages can always be ignored; back-bdb always retries when
it hits a deadlock.
--
-- Howard Chu
CTO, Symas Corp. http://www.symas.com
Director, Highland Sun http://highlandsun.com/hyc/
Chief Architect, OpenLDAP http://www.openldap.org/project/
3:33 a.m.
Howard Chu wrote:
Adrien Futschik wrote:
> Considering that M1& M3 are on the same server and therefore have
> exactly the
> same time, if this was a time related problem, I should'nt get any
> "CSN too
> old" messages between M1&M3 and M2&M4, should I ?
>
> I have also noticed that when M1 gets a new entry and passes it to
> M2&M3&M4,
> when M2&M3&M4 revieve it, they also pass it to M2&M3&M4 ! I don't
> understand
> why this happends but it look's very much like this is what's happening,
> because sometimes, M2 would have passed-it to M4, before M4 has actualy
> revieved the add order from M1.
>
> I therefore happend to notice that sometimes, entries send from M1 are
> revieved in the wrong ordrer by other masters and therefore some
> entries may
> be skipped !!!
Yes, that makes sense. The CSN check assumes changes will always be
received in the same order they were sent from the provider. Obviously
in this case this assumption is wrong. You should submit an ITS for this.
This problem was discussed on the -devel list back in 2007; the code
ought to be using a spanning tree/routing algorithm to ensure that when
multiple routes exist for propagating a change, the change is delivered
exactly once. Unfortunately no one has spent any further time on this
issue since then.
But the CSN is supposed to guarantee that regardless of the order
servers converge to the same status. In fact, if entries are received
in different order but carry an entryCSN attribute that is newer, the
newer should take place (and be propagated further through
slapo-syncprov if MMR); if identical or older, should be ignored (and
not propagated). If the modification that comes in implies something
odd like a missing parent or so, glue entries should be created, to be
replaced by the right entry as soon as it comes in. In MMR, assuming
perfect symmetry, we could do something like ignoring entries that come
from a provider with the entryCSN generated by another provider, under
the assumption we will eventually get it from the right provider? Or
better (and symmetrical) do not propagate entries whose CSN was not
generated by ourselves, under the assumption the one that generated the
CSN will propagate it?
p.
Ing. Pierangelo Masarati
OpenLDAP Core Team
SysNet s.r.l.
via Dossi, 8 - 27100 Pavia - ITALIA
http://www.sys-net.it
-----------------------------------
Office: +39 02 23998309
Mobile: +39 333 4963172
Fax: +39 0382 476497
Email: ando(a)sys-net.it
-----------------------------------
3:47 a.m.
Pierangelo Masarati wrote:
Howard Chu wrote:
> Adrien Futschik wrote:
>> Considering that M1& M3 are on the same server and therefore have
>> exactly the
>> same time, if this was a time related problem, I should'nt get any
>> "CSN too
>> old" messages between M1&M3 and M2&M4, should I ?
>>
>> I have also noticed that when M1 gets a new entry and passes it to
>> M2&M3&M4,
>> when M2&M3&M4 revieve it, they also pass it to M2&M3&M4 ! I
don't
>> understand
>> why this happends but it look's very much like this is what's happening,
>> because sometimes, M2 would have passed-it to M4, before M4 has actualy
>> revieved the add order from M1.
>>
>> I therefore happend to notice that sometimes, entries send from M1 are
>> revieved in the wrong ordrer by other masters and therefore some
>> entries may
>> be skipped !!!
> Yes, that makes sense. The CSN check assumes changes will always be
> received in the same order they were sent from the provider. Obviously
> in this case this assumption is wrong. You should submit an ITS for this.
>
> This problem was discussed on the -devel list back in 2007; the code
> ought to be using a spanning tree/routing algorithm to ensure that when
> multiple routes exist for propagating a change, the change is delivered
> exactly once. Unfortunately no one has spent any further time on this
> issue since then.
But the CSN is supposed to guarantee that regardless of the order
servers converge to the same status. In fact, if entries are received
in different order but carry an entryCSN attribute that is newer, the
newer should take place (and be propagated further through
slapo-syncprov if MMR); if identical or older, should be ignored (and
not propagated). If the modification that comes in implies something
odd like a missing parent or so, glue entries should be created, to be
replaced by the right entry as soon as it comes in.
Right, but you're talking about the entryCSN of a replicated entry, and not
the CSN that was sent in the sync cookie. The two don't have to be the same,
particularly if there are a lot of writes active on the provider.
When a consumer accepts an out of order cookie CSN then any other consumers
cascaded off it will receive incomplete data. (The consumer claims to be up to
date as of revision X, but it is in fact missing revision X-1.)
In MMR, assuming
perfect symmetry, we could do something like ignoring entries that come
from a provider with the entryCSN generated by another provider, under
the assumption we will eventually get it from the right provider? Or
better (and symmetrical) do not propagate entries whose CSN was not
generated by ourselves, under the assumption the one that generated the
CSN will propagate it?
That assumes a fully connected star topology. Such a layout won't scale; the
intention is for this to work even with irregular topologies. E.g.
A - B G
| | | \
C - D - E - F - H
--
-- Howard Chu
CTO, Symas Corp. http://www.symas.com
Director, Highland Sun http://highlandsun.com/hyc/
Chief Architect, OpenLDAP http://www.openldap.org/project/
11:33 p.m.
Le lundi 02 mars 2009 12:18:23, Howard Chu a écrit :
Adrien Futschik wrote:
> Considering that M1& M3 are on the same server and therefore have
> exactly the same time, if this was a time related problem, I should'nt
> get any "CSN too old" messages between M1&M3 and M2&M4, should I
?
>
> I have also noticed that when M1 gets a new entry and passes it to
> M2&M3&M4, when M2&M3&M4 revieve it, they also pass it to
M2&M3&M4 ! I
> don't understand why this happends but it look's very much like this is
> what's happening, because sometimes, M2 would have passed-it to M4,
> before M4 has actualy revieved the add order from M1.
>
> I therefore happend to notice that sometimes, entries send from M1 are
> revieved in the wrong ordrer by other masters and therefore some entries
> may be skipped !!!
Yes, that makes sense. The CSN check assumes changes will always be
received in the same order they were sent from the provider. Obviously in
this case this assumption is wrong. You should submit an ITS for this.
This problem was discussed on the -devel list back in 2007; the code ought
to be using a spanning tree/routing algorithm to ensure that when multiple
routes exist for propagating a change, the change is delivered exactly
once. Unfortunately no one has spent any further time on this issue since
then.
I am not whethere it is M1 that sends them in the wrong order to M2 and then
cascaded to M3&M4, or if it is the order of M2 queue's that's wrong. I guess
this must be the second option.
I'll submit an ITS right away.
Personnaly I believe that the best way to avoid this problem to happen, would
be not to propagate entries just recieved from an other master.
Adrien Futschik
4662
days inactive
4666
days old
openldap-technical@openldap.org
8 comments
5 participants
participants (5)
-
a.schild＠aarboard.ch -
Adrien Futschik -
Howard Chu -
Pierangelo Masarati -
Quanah Gibson-Mount