June 15, 201313 yr My drives were Unformatted therefore there was no filesystem housekeeping occurring on the superblock. I'm guessing that if you have some highly filled drives, any form of superblock update occurs in multiple places. Then there is journaling (which also gets applied to/from the parity drive also). This is one of the reasons I had a caching controller for parity. Any journaling IO on the parity disk would have been cached. Tom's statement about block I/O vs logical I/O makes sense. All my drives are on the same hardware, therefore block I/O may be equal for all drives. There seems to be a number of factors coming into play here.
June 16, 201313 yr Author All of my drives are on the same hardware as well, although they are formatted and in the array. What kind of counts do you get when you run a parity check against your active array?
June 16, 201313 yr All of my drives are on the same hardware as well, although they are formatted and in the array. What kind of counts do you get when you run a parity check against your active array? I have not looked at it that closely. Plus my arrays are fairly new. It takes hours right now. I'll have to look into it later after I run through a different battery of tests.
June 16, 201313 yr Author All of my drives are on the same hardware as well, although they are formatted and in the array. What kind of counts do you get when you run a parity check against your active array? I have not looked at it that closely. Plus my arrays are fairly new. It takes hours right now. I'll have to look into it later after I run through a different battery of tests. No rush ... but if, the next time you run a parity check, you'd (a) clear stats; then (b) run the check; and finally © post the read counters it'd be interesting to know if your behavior is different.
Archived
This topic is now archived and is closed to further replies.