Hi, On 8/5/2015 4:48 PM, Mathivanan Naickan Palanivelu wrote: > Iam not sure if you are alluding that it could be a flow control issue > in checkpoint service? > Also, you have increased the MDS_SOCK_SND_RCV_BUF_SIZE=33554432, > that's a very big value, > and i don't know if there's any benchmark data associated with that size! > Mahesh could comment on this.
MDS_SOCK_SND_RCV_BUF_SIZE is translated in to SO_SNDBUF setsockopt() option value , we may find benchmark in man page "setsockopt(tcp_cb->DBSRsock, SOL_SOCKET, SO_SNDBUF, &sndbuf_size, sizeof(sndbuf_size)" ====================================================================================================================== SO_SNDBUF Sets or gets the maximum socket send buffer in bytes. The kernel doubles this value (to allow space for bookkeeping overhead) when it is set using setsockopt(2), and this doubled value is returned by getsockopt(2). The default value is set by the /proc/sys/net/core/wmem_default file and the maximum allowed value is set by the /proc/sys/net/core/wmem_max file. The minimum (doubled) value for this option is 2048. ====================================================================================================================== -AVM On 8/5/2015 4:48 PM, Mathivanan Naickan Palanivelu wrote: > Iam not sure if you are alluding that it could be a flow control issue in > checkpoint service? > Also, you have increased the MDS_SOCK_SND_RCV_BUF_SIZE=33554432, that's a > very big value, > and i don't know if there's any benchmark data associated with that size! > Mahesh could comment on this. > > A question i have is, did your exercise succeed when using the default buffer > sizes for > MDS_SOCK_SND_RCV_BUF_SIZE and DTM_SOCK_SND_RCV_BUF_SIZE? > > > Thanks, > Mathi. > > ----- [email protected] wrote: > >> Hi Mathi, >> >> 1285 is different issue altogether, this is bug in checkpoint >> service. One >> of the frame gets dropped and rest of the frames are interpreted as >> stale >> messages. >> >> Regards, >> Girish >> >> -----Original Message----- >> From: Mathivanan Naickan Palanivelu [mailto:[email protected]] >> Sent: Wednesday, August 05, 2015 3:13 PM >> To: [email protected] >> Cc: [email protected] >> Subject: Re: [users] MDS receives stale message >> >> Hi, >> >> Ticket https://sourceforge.net/p/opensaf/tickets/1285/ exists for >> this >> issue, now that you have reported it again on 4.6, we shall prioritize >> this >> for the next upcoming release. >> >> Mathi. >> >> >> ----- [email protected] wrote: >> >>> Hi, >>> >>> Issue seen in opensaf 4.6 also >>> >>> Setup/environment details: >>> >>> Linux: Fedora 22, Linux localhost.localdomain 4.0.4-301.fc22.i686 >> #1 >>> SMP Thu May 21 13:43:18 UTC 2015 i686 i686 i386 GNU/Linux Opensaf >>> version: 4.6 [root@localhost sbin]# systemctl --version systemd 219 >>> +PAM +AUDIT +SELINUX +IMA -APPARMOR +SMACK +SYSVINIT +UTMP >>> +LIBCRYPTSETUP GCRYPT +GNUTLS +ACL +XZ -LZ4 +SECCOMP +BLKID >> +ELFUTILS >>> ++KMOD +IDN >>> >>> Also I exported export MDS_SOCK_SND_RCV_BUF_SIZE=33554432 and set >>> DTM_SOCK_SND_RCV_BUF_SIZE=33554432 in /etc/opensaf/dtmd.conf >>> >>> Modified checkpoint application attached. >>> >>> Steps: >>> 1. start opensaf on active and standby node 2. start ckpt_demo on >>> active, ./ckpt_demo 1 3. start ckpt_demo on standby, ./ckpt_demo 0 >>> >>> Notice error message in mds.log on standby node >>> >>> Please provide fix for this defect. >>> >>> Regards, >>> Girish >>> >>> -----Original Message----- >>> From: Mathivanan Naickan Palanivelu >> [mailto:[email protected]] >>> Sent: Thursday, May 07, 2015 7:59 PM >>> To: [email protected] >>> Cc: [email protected] >>> Subject: Re: [users] MDS receives stale message >>> >>> Hi, >>> >>> There have been some fixes done after 4.3, I think you should try >> this >>> on >>> 4.6 and let us know if the issue is still seen? >>> >>> One ticket worth mentioning that you might want to look at related >> to >>> buffer size increase and fragmentation handling would be >>> http://sourceforge.net/p/opensaf/tickets/654/. >>> >>> >>> Thanks, >>> Mathi. >>> >>> ----- [email protected] wrote: >>> >>>> Hi, >>>> >>>> >>>> >>>> We use opensaf 4.3.1 with TCP as transport, in active-standby >>> setup. >>>> >>>> >>>> When we push lot of data for checkpointing from active, we get >>> error >>>> message in mds.log on standby >>>> >>>> >>>> >>>> MDTM: Some stale message recd, hence dropping adest= >>>> >>>> >>>> >>>> Any idea what went wrong? Is there any fix already available for >>> this >>>> issue in later releases? >>>> >>>> >>>> >>>> Regards, >>>> >>>> Girish >>>> >>>> -- >>>> . >>>> >> ---------------------------------------------------------------------- >>>> -------- One dashboard for servers and applications across >>>> Physical-Virtual-Cloud Widest out-of-the-box monitoring support >>> with >>>> 50+ applications Performance metrics, stats and reports that give >>> you >>>> Actionable Insights Deep dive visibility with transaction tracing >>>> using APM Insight. >>>> http://ad.doubleclick.net/ddm/clk/290420510;117567292;y >>>> _______________________________________________ >>>> Opensaf-users mailing list >>>> [email protected] >>>> https://lists.sourceforge.net/lists/listinfo/opensaf-users >>> -- >>> . >> -- >> . > ------------------------------------------------------------------------------ > _______________________________________________ > Opensaf-users mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/opensaf-users ------------------------------------------------------------------------------ _______________________________________________ Opensaf-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/opensaf-users
