And how much ram was enough?
Im using v2/udp nfs and have disabled v3 tcp but the performance on large files stinks. Are you using a solaris machine? Im wondering if its a OS specific problem or a Netapp config problem. Everything so far points at the 630.
--- On Wed, 18 Mar 1998 18:10:53 -0800 (PST) Jaye Mathisen mrcpu@cdsnet.net wrote:
Yes, I am, works fabulous once I got enough RAM in that baby.
I find best results with v2/UDP NFS, and disable v3 and TCP mounts completely.
Also, trim your misc.jobs.offered, and control directories regularly, large dirs still kill the Netapp until they get their Btree directory implementation running.
On Wed, 18 Mar 1998, David Power wrote:
David Power V-P Operations Insync Internet Services Inc.
+--- In a previous state of mind, David Power dpower@insync.net wrote: | | And how much ram was enough? | | Im using v2/udp nfs and have disabled v3 tcp but the performance on large | files stinks. Are you using a solaris machine? Im wondering if its a OS | specific problem or a Netapp config problem. Everything so far points at | the 630.
This sounds very odd. We used NetApp's for news at my previous job and they worked beautifully.
What OS and patch levels do you have on the Sun. Not that SS20's are real speed demons...
What version of DataONTAP do you have on the filer? What ndd setting have you applied to the SS20?
Doe the large writes correspond to a specific news hierarchy (ie: misc.jobs.* or alt.binaries.*)?
What does sysstat look like when this occurs?
Alexei
Hmmm, I have both my filers maxed out. 256 wasn't enough in the 540, but 512 in the 630 has been fine.
I'm using a FreeBSD box.
With iozone on full-duplex 100mbit, I get about 8MB/second sustained for as large a file as I want...
On Wed, 18 Mar 1998, David Power wrote:
And how much ram was enough?
Im using v2/udp nfs and have disabled v3 tcp but the performance on large files stinks. Are you using a solaris machine? Im wondering if its a OS specific problem or a Netapp config problem. Everything so far points at the 630.
--- On Wed, 18 Mar 1998 18:10:53 -0800 (PST) Jaye Mathisen mrcpu@cdsnet.net wrote:
Yes, I am, works fabulous once I got enough RAM in that baby.
I find best results with v2/UDP NFS, and disable v3 and TCP mounts completely.
Also, trim your misc.jobs.offered, and control directories regularly, large dirs still kill the Netapp until they get their Btree directory implementation running.
On Wed, 18 Mar 1998, David Power wrote:
David Power V-P Operations Insync Internet Services Inc.
Also, how are your netstat -i counters doing? I had a problem with reliable full-duplex vs. half-duplex detection, and am actually getting better overall throughput via half-duplex. The system just feels faster, although some of the benchmark numbers drop.
Is the filer doing anything besides news? I find that even with my 630, cache ages are only in the few minutes range, if you're using it for a variety of services, it may be unable to cache effectively.
On Wed, 18 Mar 1998, David Power wrote:
And how much ram was enough?
Im using v2/udp nfs and have disabled v3 tcp but the performance on large files stinks. Are you using a solaris machine? Im wondering if its a OS specific problem or a Netapp config problem. Everything so far points at the 630.
--- On Wed, 18 Mar 1998 18:10:53 -0800 (PST) Jaye Mathisen mrcpu@cdsnet.net wrote:
Yes, I am, works fabulous once I got enough RAM in that baby.
I find best results with v2/UDP NFS, and disable v3 and TCP mounts completely.
Also, trim your misc.jobs.offered, and control directories regularly, large dirs still kill the Netapp until they get their Btree directory implementation running.
On Wed, 18 Mar 1998, David Power wrote:
David Power V-P Operations Insync Internet Services Inc.
-
Alexei Rodriguez -
David Power -
Jaye Mathisen