Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
/ 2004-09-30 13:56:05 -0400
\ Dan Didier:
> First some background on our systems. They are both Dell 2400 with a
> built in raid controller with 64megs of memory. The OS is on a separate
> mirrored RAID partition. The data is a RAID 5 partition. We are using
> 100mb Intel NICs for data sync. Dual processor 733Mhz. Both are
> running Redhat9 kernel 2.4.20-31.9smp. DRBD 0.7.4
>
> I have been watching our systems and have some questions that hopefully
> some of you can help with.
> My first question is what is the difference between speed and K/sec. As
> you can see below the first output shows:
> finish: 26:02:30 speed: 508 (1,428) K/sec, while the second shows:
> finish: 1:17:08 speed: 10,632 (1,576) K/sec
^ current `average since sync started
note that the "," marks the thousand, not the decimal point
> The K/sec is the same but the speed is much lower. Is this because of
> disk write issues?
its only a rough guestimate anyways.
> On my secondary box I get the follwing in my dmesg:
> aacraid:ID(0:02:0) Timeout detected on cmd[0x2a]
> aacraid:SCSI Channel[0]: Timeout Detected On 1 Command(s)
> aacraid:ID(0:02:0) Timeout detected on cmd[0x2a]
> aacraid:SCSI Channel[0]: Timeout Detected On 1 Command(s)
> aacraid:ID(0:04:0) Timeout detected on cmd[0x28]
> aacraid:SCSI Channel[0]: Timeout Detected On 1 Command(s)
> aacraid:ID(0:03:0) Timeout detected on cmd[0x2a]
> aacraid:SCSI Channel[0]: Timeout Detected On 1 Command(s)
> aacraid:ID(0:04:0) Timeout detected on cmd[0x2a]
> aacraid:SCSI Channel[0]: Timeout Detected On 1 Command(s)
> aacraid:ID(0:02:0) Timeout detected on cmd[0x2a]
> aacraid:SCSI Channel[0]: Timeout Detected On 1 Command(s)
sorry. not drbds fault.
it only stresses your hardware more than your hardware seems to like.
this of course will bring down your performance big time.
> I have two identical boxes. I have swapped the drives out to make sure
> this isn't a hardware issue. This always occurs on the secondary which
> I suppose would make sense as that is where all the writes are occuring.
hm. maybe aacraid driver issue. dunno.
Lars Ellenberg
--
please use the "List-Reply" function of your email client.