Re: IDS 7.31 UD1 vs Solaris 8
Thank you to all those who made suggestions (and apologies for the late response - "not waving but drowning" :-^ ) We ended up doubling the CKPTINVL value and reducing the LRU_MAX_DIRTY and LRU_MIN_DIRTY by a factor of 10 - not ideal I know but it makes the perception of the system to users a lot less 'lumpy' and checkpoint times are down to 20 - 30 seconds. I still convinced that the underlying problem is still there tho'. I've checked - as far as I can - the fragmentation of the COOKED chunk file (cf ArtK) (but it was created using 'mkfile ...' which writes ZEROs to the whole file and, I assume, tries to make with the least fragmentation as possible). I have compared /etc/system files and they are all the same (cf NormaJeanS) - lots of semaphores and unlimited shared memory. IDS uses approx 35% of physical memory. I checked swap and swapping (cf SamG) and everything seems OK - and 'iostat -xpn' is indeed a source of useful data. Two further points of interest a) looking at the output from 'sar -u' there is very little idle time but a lot of WIO time b) with the changed onconfig parameters the write cache figures are down to 50% TIA -- Mark Thornber ========================= E M Thornber CEng MIEE Enchanted Systems Limited Software Toolsmiths +44 (0) 1503 272097
"a lot of WIO time" is that "waiting for I/O"? hmmmm... do you have an i/o controller bottleneck? Norma Jean -----Original Message----- From: emthornber@iee.org [mailto:emthornber@iee.org] Sent: Sunday, April 25, 2004 12:57 PM To: ids@iiug.org; forum.subscriber@iiug.org Subject: Re: IDS 7.31 UD1 vs Solaris 8 [2882] Thank you to all those who made suggestions (and apologies for the late response - "not waving but drowning" :-^ ) We ended up doubling the CKPTINVL value and reducing the LRU_MAX_DIRTY and LRU_MIN_DIRTY by a factor of 10 - not ideal I know but it makes the perception of the system to users a lot less 'lumpy' and checkpoint times are down to 20 - 30 seconds. I still convinced that the underlying problem is still there tho'. I've checked - as far as I can - the fragmentation of the COOKED chunk file (cf ArtK) (but it was created using 'mkfile ...' which writes ZEROs to the whole file and, I assume, tries to make with the least fragmentation as possible). I have compared /etc/system files and they are all the same (cf NormaJeanS) - lots of semaphores and unlimited shared memory. IDS uses approx 35% of physical memory. I checked swap and swapping (cf SamG) and everything seems OK - and 'iostat -xpn' is indeed a source of useful data. Two further points of interest a) looking at the output from 'sar -u' there is very little idle time but a lot of WIO time b) with the changed onconfig parameters the write cache figures are down to 50% TIA -- Mark Thornber ========================= E M Thornber CEng MIEE Enchanted Systems Limited Software Toolsmiths +44 (0) 1503 272097 ----------------------------------------- ============================================================ The information contained in this message may be privileged and confidential and protected from disclosure. If the reader of this message is not the intended recipient, or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any reproduction, dissemination or distribution of this communication is strictly prohibited. If you have received this communication in error, please notify us immediately by replying to the message and deleting it from your computer. Thank you. Tellabs ============================================================