Re: I/O write () chunk 1

In article <5mvavm$jg8@cssun.mathcs.emory.edu>, Croft, Evelyn <croft.evelyn@dorseylaw.com> wrote: >Hello Everyone, > >Can anyone tell me the cause/meaning of the following problem. > >I/O error 'read': expect 2048 actual -1 addr 0x16dc0 errno 5 retrys 0 >I/O error 'read': expect 2048 actual -1 addr 0x16dc0 errno 5 retrys 1 >I/O error, Primary Chunk '/dev/rootprime' -- offline Yikes! After all these years, they still use the same diagnostic message format that I put in as a "quick fix" in OnLine 4.0 seven (?) years ago?! [-: >I/O () chunk 1, pagenum 93632, pagecnt 1 --> errno =3D 5 >INFORMIX-OnLine Must Abort Translation: OnLine tried to do a 2K (one page) read, and the operating system returned a -1 with errno == 5, implying an i/o error. It retried the operation once (the default; it's tuneable) but got the same error. All bets are off at that point; to avoid further damage, OnLine aborts. (By the way, OnLine admins out there, it's a good idea to watch for such messages even if the retry IS successful... it could indicate an impending hardware failure -or- a hardware compatibility problem that only shows under heavy loads.) >Root chunk down. >INFORMIX-OnLine entering ABORT mode!!! >-- Online Aborting -- us=3Dc0011e60, pid=3D23437, uid=3D219 >-- Online Aborting -- us=3Dc000c14c, pid=3D238, uid=3D0 >I/O error 'write': expect 32768 actual -1 addr 0x24a1 errno 5 retrys 0 >I/O error 'write': expect 32768 actual -1 addr 0x24a1 errno 5 retrys 1 >I/O write () chunk 1, pagenum 9377, pagecnt 16 --> errno =3D 5 As it was aborting, it tried to write 16 pages (perhaps checkpoint or shutdown info to the log) and it again got i/o errors from the operating system. >INFORMIX-OnLIne Must ABORT >Root chunk down. >-- Online Aborting -- us=3Dc000f960, pid=3D26172, uid=3D562 >INFORMIX-OnLine Stopped > >This is the second time in the last 6 months that the primary >chunk has gone down. Also any suggestions in preventing this >problem is MUCH appreciated. Replace the disk in question (the one that contains the root chunk). Make sure all disk hardware you are using is certified for your O/S. (I've seen this happen before where the disk hardware or i/o hardware or firmware was mismatched with the O/S.) It's also possible that i/o errors were noted by the operating system and recorded in a system log somewhere. >Thx. >Eve >Croft.Evelyn@dorseylaw.com -- Alan Denney aland@informix.com yosemite@netcom.com Informix Platform Engineering "I believe there are more instances of the abridgment of the freedom of the people by gradual and silent encroachment of those in power than by violent and sudden usurpations." -- James Madison