Do you have a clear plan to remove this requirement? We want to know if we can't change node_id due to our architecture, when we could get a no-this-limit release to upgrade? After all, our products have been deployed to many customers so we have to think about upgrade and compatibility issues.
Thanks, Jianfeng -----Original Message----- From: Anders Widell [mailto:anders.wid...@ericsson.com] Sent: Tuesday, October 11, 2016 4:10 PM To: Jianfeng Dong <jd...@juniper.net>; Neelakanta Reddy <reddy.neelaka...@oracle.com>; opensaf-users@lists.sourceforge.net Subject: Re: [users] OpenSAF release 5.0.1 can not promote SC after enable "headless cluster" feature Yes, this is required with the current implementation. It might be possible to remove this requirement - I will think about how it can be done. regards, Anders Widell On 10/11/2016 09:06 AM, Jianfeng Dong wrote: > Is it obligatory that controller must have a slower slot_id than payload if > we want to enable "headless" feature? > If it is obligatory, seems it's a big change to our architecture, but I will > have a try at least. > > Thanks, > Jianfeng > > -----Original Message----- > From: Anders Widell [mailto:anders.wid...@ericsson.com] > Sent: Tuesday, October 11, 2016 2:30 PM > To: Jianfeng Dong <jd...@juniper.net>; Neelakanta Reddy > <reddy.neelaka...@oracle.com>; opensaf-users@lists.sourceforge.net > Subject: Re: [users] OpenSAF release 5.0.1 can not promote SC after > enable "headless cluster" feature > > There is a one-to-one mapping between /etc/opensaf/slot_id and the node_id. > Simply make sure that all your system controller nodes have lower slot_id > than any of your payloads. This file is read when the node is booted. You > should be able to do an in-service renumbering of your nodes - just be > careful so that you never have two nodes with the same node_id at the same > time. > > Yes, the assumption is there in 5.1.0 as well. > > regards, > > Anders Widell > > > On 10/11/2016 04:29 AM, Jianfeng Dong wrote: >> Yes, in our product payload's node_id is lower than SC, could you please >> tell us how to configure it? >> >> And, does this assumption exist in OpenSAF 5.1.0 as well? >> >> Thanks, >> Jianfeng >> >> -----Original Message----- >> From: Anders Widell [mailto:anders.wid...@ericsson.com] >> Sent: Tuesday, October 11, 2016 12:55 AM >> To: Jianfeng Dong <jd...@juniper.net>; Neelakanta Reddy >> <reddy.neelaka...@oracle.com>; opensaf-users@lists.sourceforge.net >> Subject: Re: [users] OpenSAF release 5.0.1 can not promote SC after >> enable "headless cluster" feature >> >> There is a (probably not so well documented :-) assumption that the system >> controllers are configured with a lower node_id than the payloads. From what >> I can see in the logs you sent, I think it looks like you have configured >> the payload with a lower node_id than the system controllers. >> >> By the way, the headless feature has been improved in OpenSAF 5.1.0 so I >> would suggest that you upgrade to that version if possible. >> >> regards, >> >> Anders Widell >> >> >> On 10/10/2016 06:04 PM, Jianfeng Dong wrote: >>> I tried with sufficient drive space but got same result, neither of the two >>> SCs can be promoted to be controller until the payload reboot. >>> >>> I also checked the network link between SC and payload, they can PING each >>> other when this issue happened. I suspect too the problem is caused by >>> IMMD/IMMND link among those nodes, but don't know how to prove it. >>> >>> From: Neelakanta Reddy [mailto:reddy.neelaka...@oracle.com] >>> Sent: Monday, October 10, 2016 8:39 PM >>> To: Jianfeng Dong <jd...@juniper.net>; >>> opensaf-users@lists.sourceforge.net >>> Subject: Re: [users] OpenSAF release 5.0.1 can not promote SC after >>> enable "headless cluster" feature >>> >>> Hi, >>> >>> Once after the "Headless" if any of the controller started then the IMMND >>> from the payaload will send the intro message to IMMD. >>> Looks like this did not happen, the following is the log from the payload: >>> >>> 2016-10-10T11:09:18.507851+08:00 pld0101 osafimmnd[3141]: message >>> repeated 2 times: [ logtrace: write failed, No space left on device] >>> 2016-10-10T11:09:18.507883+08:00 pld0101 osafimmnd[3141]: NO >>> Re-introduce-me highestProcessed:23839 highestReceived:23839 >>> 2016-10-10T11:09:18.508011+08:00 pld0101 osafimmnd[3141]: logtrace: >>> write failed, No space left on device >>> 2016-10-10T11:09:18.508129+08:00 pld0101 osafimmnd[3141]: logtrace: >>> write failed, No space left on device >>> 2016-10-10T11:09:18.508501+08:00 pld0101 osafimmnd[3141]: WA MDS >>> Send Failed to service:IMMD rc:2 >>> >>> >>> Retry, again with the sufficient space in payload. >>> >>> /Neel. >>> >>> On 2016/10/10 03:59 PM, Jianfeng Dong wrote: >>> >>> Hi, >>> >>> >>> >>> For several years we use OpenSAF(4.5.2 now) to provide HA service in our >>> product(including 2 SC and several payload cards), but our customer keep on >>> requiring that it's better to do NOT reboot payload card even if both SC >>> reload or hang. >>> >>> >>> >>> We just knew that the new release 5.0.0 has provided this feature(i.e. >>> "headless cluster"), so we installed 5.0.0 into our product and enable >>> "headless" feature by setting "IMMSV_SC_ABSENCE_ALLOWED" to 900 seconds. >>> After installation we found it worked fine, our system with new OpenSAF >>> release can start to run successfully, all SC and payload cards can be >>> "UP", and payload card will NOT reboot immediately after we reload both SC. >>> >>> >>> >>> However we got a problem that, neither of two SC can't be promoted to be >>> controller after reboot until the "headless" payload reboot due to >>> 'IMMSV_SC_ABSENCE_ALLOWED' timeout after 900 seconds. Seems OpenSAF modules >>> in both SC just wait there and do nothing, till payload reboot due to >>> timeout, then OpenSAF in SC continue to run, whole system recovered finally. >>> >>> >>> >>> We thought ticket #1828 may has resolved this issue so we took another try >>> with release 5.0.1 but got same result. >>> >>> >>> >>> Could you please tell us in our case, why OpenSAF in both SC could not run >>> until payload card(in "headless" status) rebooted due to timeout? >>> >>> Besides 'IMMSV_SC_ABSENCE_ALLOWED', is there any other variable or >>> parameter need to set/modify to enable 'headless cluster' feature? Do we >>> miss anything? >>> >>> Attachments are the syslog of SC and payload card when this problem >>> happened, hope the log files can help us to find out the root cause. >>> >>> >>> >>> Much appreciated to any comment, thanks! >>> >>> >>> >>> Regards, >>> >>> Jianfeng Dong >>> >>> >>> >>> >>> >>> >>> -------------------------------------------------------------------- >>> - >>> - >>> -------- >>> >>> Check out the vibrant tech community on one of the world's most >>> >>> engaging tech sites, SlashDot.org! http://sdm.link/slashdot >>> >>> >>> >>> >>> _______________________________________________ >>> >>> Opensaf-users mailing list >>> >>> Opensaf-users@lists.sourceforge.net<mailto:Opensaf-users@lists.sourc >>> e >>> f >>> orge.net> >>> >>> https://lists.sourceforge.net/lists/listinfo/opensaf-users >>> >>> -------------------------------------------------------------------- >>> - >>> - >>> -------- Check out the vibrant tech community on one of the world's >>> most engaging tech sites, SlashDot.org! http://sdm.link/slashdot >>> _______________________________________________ >>> Opensaf-users mailing list >>> Opensaf-users@lists.sourceforge.net >>> https://lists.sourceforge.net/lists/listinfo/opensaf-users >>> > ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot _______________________________________________ Opensaf-users mailing list Opensaf-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/opensaf-users