Geoffrey, > We have used AVS 4.0 (SNDR) to replicate the partition in two server. > We got the following problems: > > after completing the full sync and the state of SNDR turned to > "replicating", > but the "Percentage of volume requiring sync" (pct) value does not > equal to zero. > > [EMAIL PROTECTED] # dsstat > name t s pct role ckps dkps tps svt > /dev/md/rdsk/d8 P R 0.02 net - Inf 4294967295 NaN > /dev/md/rdsk/d6 bmp 0 0 0 0 > /dev/md/rdsk/d9 P R 0.20 net - 0 0 0 > /dev/md/rdsk/d7 bmp 0 0 0 0 > > Is there any problem with my application?
The maintenance of the "number of bits set in a bitmap", represented in 'dsstat' as a percentage of all bits available, is not done by counting each "1" or "0", on each pass of collecting kernel kstat data. Instead this value is based on incrementing and decrementing a counter representing all the bits in a bitmap. Since there are various places within SNDR where kstat counters are maintained, it is done without locking, so as to not to introduce unneeded synchronization just for maintaining statistical counters. Momentarily placing a replica in logging mode, then a fast resync, bring the set back into replicating mode should clear these values. Just as an FYI, the value 4294967295 is the unsigned decimal representation of 0xFFFFFFFF, or a -1 as a signed 32-bit value. In other words, the calculation for transactions / second (tps), went negative by 1. This large positive number, and the value of 'NaN', are the result of imprecise kstat data collection. - Jim > > > Thanks > Geoffrey > > > This message posted from opensolaris.org > <dsstat.txt>_______________________________________________ > storage-discuss mailing list > [email protected] > http://mail.opensolaris.org/mailman/listinfo/storage-discuss _______________________________________________ storage-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/storage-discuss
