Linux + informix 7.30
Kernel 2.3.18 IDS 7.30UC10 I have a rather cooked linux distro, basically redhat 6.0 but rather warmed over. It's a dual celeron with 384 Meg which I use to toy with various db's (Oracle, Sybase and Informix). I recently (this morning) decided I'd rather like to throw a lot of socket connects (using esqlc) at Informix 7.3 to see how it coped. Hence I found it necessary to configure a socket setup. It doesn't work. Error 9 in the log. A bit of digging revealed that. a) My version IDS 7.30U might, shall we say, not have been the finest product ever to leave memlo. So I upgraded it to 7.30UC10 b) Informix are validating against 2.0.36 and even then with a slew of large file descriptor patches. Given the error I'm seeing means bad file number ie a descriptor error one is suspicious that my woes are due to the lack of these patches. It is necessary for me to be on a 2.3 kernel due to poor disk performance for the BX chipset on the 2.2 range. Does anyone have kernel version recommendations and does anyone know if these Informix patches are being bundled up into release kernels, if so which ones? cheers mja Sent via Deja.com http://www.deja.com/ Before you buy.
In article <8ro6uq$rdq$1@nnrp1.deja.com>, martyn.ayshford@orange.co.uk wrote: >Kernel 2.3.18 >IDS 7.30UC10 > >large file descriptor patches. Given the error I'm seeing means bad >file number ie a descriptor error one is suspicious that my woes are due >to the lack of these patches. If the errors you are seeing are:- "dupfd: fd %d, max %d\\n" This apppears to be the server trying to detect the maximum available fd. Look at the fd sequence, it looks like a binary search, TTFN -- Roger Think of the mess on the carpet. Sensible people do all their demon-summoning in the garage, which you can just hose down afterwards. -- damerell@chiark.greenend.org.uk
? Confused However the follwoing may be of use..... I decided that I'd rather like to check 9.21 over (on the basis I shall have to upgrade some rather large instances professionly) The installation and branding went smoothly enough however it would not start due to my glibc being to old. So I upgraded it. Then it would start with shared memory but not without. Same error (error 9) which means bad file number. Mmmm suspicious. So I finally bit the bullet, I rather like 2.3.18, it's been no trouble and my disks fly, and downloaded the latest stable Kernel (2.2.17) I toyed with Alan Cox's .18.pre's but discreation got the better of valour. Compiled it and booted the box. Informix 9.21UC comes on line ( with sockets) I can access my instance from my other linux box running the SDK (actually the full 7.30 shooting match). From this we can say that the descriptor patch is available in this kernel, and reading informix's 9.21 release docs we can infer that it got rolled into 2.12. Lamentably the same can not be said for BX chipset support, my hardisks are now performing dismally. Rollon 2.4. mja article <slrn8u0qsa.rpt.roger@knuth.computer-surgery.co.uk>, roger@knuth.computer-surgery.co.uk (Roger Gammans) wrote: > In article <8ro6uq$rdq$1@nnrp1.deja.com>, martyn.ayshford@orange.co.uk wrote: > >Kernel 2.3.18 > >IDS 7.30UC10 > > > >large file descriptor patches. Given the error I'm seeing means bad > >file number ie a descriptor error one is suspicious that my woes are due > >to the lack of these patches. > > If the errors you are seeing are:- > "dupfd: fd %d, max %d\\n" > > This apppears to be the server trying to detect the maximum available > fd. Look at the fd sequence, it looks like a binary search, > > TTFN > -- > Roger > Think of the mess on the carpet. Sensible people do all their > demon-summoning in the garage, which you can just hose down afterwards. > -- damerell@chiark.greenend.org.uk > > Sent via Deja.com http://www.deja.com/ Before you buy.
2.4 pre -tests are out there, have you tried one on a test box yet?? Best test and find the bugs as early as possible. martyn.ayshford@orange.co.uk wrote in message <8rt5lm$f3i$1@nnrp1.deja.com>... >? > >Confused > >However the follwoing may be of use..... > >I decided that I'd rather like to check 9.21 over (on the basis I shall >have to upgrade some rather large instances professionly) > >The installation and branding went smoothly enough however it would not >start due to my glibc being to old. So I upgraded it. Then it would >start with shared memory but not without. Same error (error 9) which >means bad file number. Mmmm suspicious. > >So I finally bit the bullet, I rather like 2.3.18, it's been no trouble >and my disks fly, and downloaded the latest stable Kernel (2.2.17) I >toyed with Alan Cox's .18.pre's but discreation got the better of >valour. > >Compiled it and booted the box. Informix 9.21UC comes on line ( with >sockets) I can access my instance from my other linux box running the >SDK (actually the full 7.30 shooting match). > >From this we can say that the descriptor patch is available in this >kernel, and reading informix's 9.21 release docs we can infer that it >got rolled into 2.12. > >Lamentably the same can not be said for BX chipset support, my hardisks >are now performing dismally. Rollon 2.4. > >mja > > > > > article <slrn8u0qsa.rpt.roger@knuth.computer-surgery.co.uk>, > roger@knuth.computer-surgery.co.uk (Roger Gammans) wrote: >> In article <8ro6uq$rdq$1@nnrp1.deja.com>, martyn.ayshford@orange.co.uk >wrote: >> >Kernel 2.3.18 >> >IDS 7.30UC10 >> > >> >large file descriptor patches. Given the error I'm seeing means bad >> >file number ie a descriptor error one is suspicious that my woes are >due >> >to the lack of these patches. >> >> If the errors you are seeing are:- >> "dupfd: fd %d, max %d\\n" >> >> This apppears to be the server trying to detect the maximum available >> fd. Look at the fd sequence, it looks like a binary search, >> >> TTFN >> -- >> Roger >> Think of the mess on the carpet. Sensible people do all their >> demon-summoning in the garage, which you can just hose down >afterwards. >> -- damerell@chiark.greenend.org.uk >> >> > > >Sent via Deja.com http://www.deja.com/ >Before you buy.
smooth1, your a bloody mindreader. 2.4.pre9 is on it's way to my hard disk at this very moment, for precisely that reason. I had a few problems with 2.2 pre's that were kindly fixed prior to 2.2 hitting the streets. Further investigation with 2.2.17 has uncovered system (my system) capatibility problems, problems that are not evident in 2.3.18, hence I'm going to get 2.4.pre9 down find out if all my system stuff works, and ensure you can bring an Informix instance on-line, if not I shall figure out why and send Linus a mail. mja In article <39e25407_3@news1.vip.uk.com>, "smooth1" <smooth1@iclway.co.uk> wrote: > 2.4 pre -tests are out there, have you tried one on a test box yet?? > > Best test and find the bugs as early as possible. > > martyn.ayshford@orange.co.uk wrote in message > <8rt5lm$f3i$1@nnrp1.deja.com>... > >? > > > >Confused > > > >However the follwoing may be of use..... > > > >I decided that I'd rather like to check 9.21 over (on the basis I shall > >have to upgrade some rather large instances professionly) > > > >The installation and branding went smoothly enough however it would not > >start due to my glibc being to old. So I upgraded it. Then it would > >start with shared memory but not without. Same error (error 9) which > >means bad file number. Mmmm suspicious. > > > >So I finally bit the bullet, I rather like 2.3.18, it's been no trouble > >and my disks fly, and downloaded the latest stable Kernel (2.2.17) I > >toyed with Alan Cox's .18.pre's but discreation got the better of > >valour. > > > >Compiled it and booted the box. Informix 9.21UC comes on line ( with > >sockets) I can access my instance from my other linux box running the > >SDK (actually the full 7.30 shooting match). > > > >From this we can say that the descriptor patch is available in this > >kernel, and reading informix's 9.21 release docs we can infer that it > >got rolled into 2.12. > > > >Lamentably the same can not be said for BX chipset support, my hardisks > >are now performing dismally. Rollon 2.4. > > > >mja > > > > > > > > > > article <slrn8u0qsa.rpt.roger@knuth.computer-surgery.co.uk>, > > roger@knuth.computer-surgery.co.uk (Roger Gammans) wrote: > >> In article <8ro6uq$rdq$1@nnrp1.deja.com>, martyn.ayshford@orange.co.uk > >wrote: > >> >Kernel 2.3.18 > >> >IDS 7.30UC10 > >> > > >> >large file descriptor patches. Given the error I'm seeing means bad > >> >file number ie a descriptor error one is suspicious that my woes are > >due > >> >to the lack of these patches. > >> > >> If the errors you are seeing are:- > >> "dupfd: fd %d, max %d\\n" > >> > >> This apppears to be the server trying to detect the maximum available > >> fd. Look at the fd sequence, it looks like a binary search, > >> > >> TTFN > >> -- > >> Roger > >> Think of the mess on the carpet. Sensible people do all their > >> demon-summoning in the garage, which you can just hose down > >afterwards. > >> -- damerell@chiark.greenend.org.uk > >> > >> > > > > > >Sent via Deja.com http://www.deja.com/ > >Before you buy. > > Sent via Deja.com http://www.deja.com/ Before you buy.
Addendum, if anyone is interested.... Informix 921UC2 comes up just fine with 2.4pre9. My disks are back to full speed and memory are correctly detected, it also started telling me about my USB Hub.... This just became my default kernel.... mja In article <8rvbq5$6a5$1@nnrp1.deja.com>, martyn.ayshford@orange.co.uk wrote: > smooth1, your a bloody mindreader. 2.4.pre9 is on it's way to my hard > disk at this very moment, for precisely that reason. > > I had a few problems with 2.2 pre's that were kindly fixed prior to 2.2 > hitting the streets. Further investigation with 2.2.17 has uncovered > system (my system) capatibility problems, problems that are not evident > in 2.3.18, hence I'm going to get 2.4.pre9 down find out if all my > system stuff works, and ensure you can bring an Informix instance > on-line, if not I shall figure out why and send Linus a mail. > > mja > > In article <39e25407_3@news1.vip.uk.com>, > "smooth1" <smooth1@iclway.co.uk> wrote: > > 2.4 pre -tests are out there, have you tried one on a test box yet?? > > > > Best test and find the bugs as early as possible. > > > > martyn.ayshford@orange.co.uk wrote in message > > <8rt5lm$f3i$1@nnrp1.deja.com>... > > >? > > > > > >Confused > > > > > >However the follwoing may be of use..... > > > > > >I decided that I'd rather like to check 9.21 over (on the basis I > shall > > >have to upgrade some rather large instances professionly) > > > > > >The installation and branding went smoothly enough however it would > not > > >start due to my glibc being to old. So I upgraded it. Then it would > > >start with shared memory but not without. Same error (error 9) which > > >means bad file number. Mmmm suspicious. > > > > > >So I finally bit the bullet, I rather like 2.3.18, it's been no > trouble > > >and my disks fly, and downloaded the latest stable Kernel (2.2.17) I > > >toyed with Alan Cox's .18.pre's but discreation got the better of > > >valour. > > > > > >Compiled it and booted the box. Informix 9.21UC comes on line ( with > > >sockets) I can access my instance from my other linux box running the > > >SDK (actually the full 7.30 shooting match). > > > > > >From this we can say that the descriptor patch is available in this > > >kernel, and reading informix's 9.21 release docs we can infer that it > > >got rolled into 2.12. > > > > > >Lamentably the same can not be said for BX chipset support, my > hardisks > > >are now performing dismally. Rollon 2.4. > > > > > >mja > > > > > > > > > > > > > > > article <slrn8u0qsa.rpt.roger@knuth.computer-surgery.co.uk>, > > > roger@knuth.computer-surgery.co.uk (Roger Gammans) wrote: > > >> In article <8ro6uq$rdq$1@nnrp1.deja.com>, > martyn.ayshford@orange.co.uk > > >wrote: > > >> >Kernel 2.3.18 > > >> >IDS 7.30UC10 > > >> > > > >> >large file descriptor patches. Given the error I'm seeing means > bad > > >> >file number ie a descriptor error one is suspicious that my woes > are > > >due > > >> >to the lack of these patches. > > >> > > >> If the errors you are seeing are:- > > >> "dupfd: fd %d, max %d\\n" > > >> > > >> This apppears to be the server trying to detect the maximum > available > > >> fd. Look at the fd sequence, it looks like a binary search, > > >> > > >> TTFN > > >> -- > > >> Roger > > >> Think of the mess on the carpet. Sensible people do all their > > >> demon-summoning in the garage, which you can just hose down > > >afterwards. > > >> -- damerell@chiark.greenend.org.uk > > >> > > >> > > > > > > > > >Sent via Deja.com http://www.deja.com/ > > >Before you buy. > > > > > > Sent via Deja.com http://www.deja.com/ > Before you buy. > Sent via Deja.com http://www.deja.com/ Before you buy.