FYI ---------- Forwarded message ---------- From: Shai Date: Mar 22, 2006 6:27 PM Subject: Re: raid5: switching cache buffer size, 512 --> 4096 To: Gordon H. Hi, I want to thank you very much for answering all my questions and I don't want you to think for one second that I don't appricitate your time... I do :) Ok, so getting them removed via the source is the solution.. but what if I don't have the source? Can you tell me what exactly I should look for and change? Thanks (again), Shai On 3/22/06, Gordon H. wrote: > On Wed, 22 Mar 2006, Shai wrote: > > > Ok, > > > > say I don't care about md1 being redone. Is there anything i can do to > > it that will fix these error messages? format again with some > > parameters? rebuild the raid? > > They asren't really errors - more informational. Which is why you should > comment them out of the source code if you don't want them... > > You aer probably taking a performance hit when they are printed though - > for 2 reasons, one the overhead of actually printing them (and logging > them), and the other when the cache is flushed, so best to stop it > happening, so look in the xfs manuals to see if you can fix the blocks > size on the journal. > > Gordon > > > > > > Shai > > > > On 3/22/06, Gordon H. wrote: > > > On Wed, 22 Mar 2006, Shai wrote: > > > > > > > I just found this email on the website but I don't know where the archives are. > > > > I'm not very good with mdadm and I don't want to mess things up to > > > > much trying things out. > > > > > > It's got nothing to do with mdadm - it's simply that xfs sometiems makes a > > > request for a block-size of 512 bytes, then a block size of 4096 bytes, > > > and theres only one block cache in the md system, so when the block size > > > changes it has to flush the cache. > > > > > > > Can you show me the archives? > > > > > > http://www.google.co.uk/search?hl=en&q=linux-raid+archive&btnG=Google+Search&meta= > > > > > > > Can you maybe help me out in finding how to fix it rather then comment > > > > out the error messages? > > > > > > If you can fix it, then the linux-raid crowd will welcome you with open > > > arms :) > > > > > > But I'd start with looking at the man pages for xfs and seeing if theres a > > > way to fix the block size of the journal. > > > > > > Gordon > > > > > > > > > > Thanks in advance!!! > > > > > > > > Shai > > > > > > > > On 3/22/06, Gordon H. wrote: > > > > > On Wed, 22 Mar 2006, Shai wrote: > > > > > > > > > > > Hi, > > > > > > > > > > > > I have two raid5 MDs: /dev/md0 and /dev/md1; > > > > > > > > > > > > I had broken md0 the other day and had to rebuild it. > > > > > > i've formated it as xfs and wanted to make md1 also xfs so I decided to > > > > > > move all the data from md1 to md0. > > > > > > > > > > > > while doing the cp of all that data, I'm getting lots of messages on > > > > > > the console screen which don't enable me to work on it ( i use telnet > > > > > > now ): > > > > > > > > > > > > > raid5: switching cache buffer size, 512 --> 4096 > > > > > > > raid5: switching cache buffer size, 4096 --> 512 > > > > > > > > > > > > Can someone help me resolve this? > > > > > > > > > > Search the archives - I had this many moons ago. It's something to do with > > > > > xfs using a different block-size for the journal to the data, and I think > > > > > there are mount/mkfs options to remedy this, or the other solution is to > > > > > find the kprintf in the driver and comment it out... > > > > > > > > > > (I stopped uing xfs, but not for other reasons. don't see it under ext3) > > > > > > > > > > Gordon - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
