MySQL unable to bind
We upgraded MySQL server from 4.1 to 5.0 in 3 servers. For various reasons (the most important is strict control of the versions) we always (since a lot of years) compiled the program from the sources. No problem at all with 2 of the servers, but we are unable to start the new server on the third PC, it always says that the address is already in use! Here is the exact error: 090830 10:08:16 [ERROR] Can't start server: Bind on TCP/IP port: Address already in use 090830 10:08:16 [ERROR] Do you already have another mysqld server running on port: 3306 ? Obviously there is no other mysql server running! We stop it before installing the new version. Moreover, if we reinstall the 4.1 version it starts without any problem. I tried it tens of times! We use the exactly the same compile options on the three servers. Almost anything is the same: inetd.conf, login.conf, master.passwd mysql line, group, filesystem permissions on the logs, run and data directory... Today I have done some new experiments and found that the new server starts correctly if we let it run as root (user = root)! But obviously I don't want to let it run as root... So, the question is: what could make a normal (non root) user be unable to bind to a port that root user can bind? As the same server works on the other two PCs, there must be some subtle difference in this PC, but I really cannot find what! Anybody have any suggestion? Thanks.
Re: Funny T22 Freezes with 4.5
Hi list, I m experiencing random freezes with few T22 fairly old laptops. I know those have crazy bios with few bugs. I tried a jump at UKC then disabled the acpi with no effect. I disabled the power mangement for cpu and pci bus with no more luke. I think you are going down the wrong path here. I tried so many things including disabling all io . I ve been amused to discover at last boot the laptop is faster than ever (no explanation yet ) but i do have a really singular dmesg... # dmesg|grep cpu cpu0: Intel Pentium III (GenuineIntel 686-class) 186 MHz cpu0: FPU,V86,DE,PSE,TSC,MSR,PAE,MCE,CX8,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,MMX,FXSR,SSE cpu0 at mainbus0: (uniprocessor) Nope this is not a troll with modified frequency :)) no, but it could be a troll with a snipped dmesg. Not at all ,i mean not in my case. As i said i disabled (from the crappy setup itself not @ukc) power management , for the cpu and pci. I forgot to say those pc are using the DC adapter not the battery i set cpu to max perf from the setup, so no speedstep issue :-/ Some (at least) PIII-based IBM ThinkPads come up with really funny processor speeds. Don't sweat it, it has nothing to do with reality (which is a bummer, 90% of the time, I wish it DID run at 200MHz for battery life it; is only when I do something with a Mozilla product that I wish it ran at full speed...plus an extra zero...) Possible, but what is awesome is that unexplainable fastness at reboot (aleady set to max perf). Well i made various test with Mozilla-FF ... it sucks so much, i saw it set the cpu usage to 90% most of the time (on powermac / tiger) but hey it s another topic :D I have a T2? laptop (t22? t23? too lazy to go look, and your snipped dmesg doesn't give me enough to figure out if yours is the same as mine (and yes, it says something about my social life that I'd probably recognize the dmesg more than the model number)). My A21 systems run fine on APM, but the T2? NEEDS to have APM disabled to use ACPI instead, otherwise the thing runs WAY too hot. I have 3 T22 (2647 model number) . I can provide the full dmesg of course (i mailed the entire one to dmesg@ as usual) IF you are having the same problem, I can easily see it locking up WHEN it got too warm. I saw it too (but i decided to remove the pcmcia cage to add another fan) So instead of trying to force APM, try forcing ACPI, see if it runs better. This stupid doesn't allow any ACPI that's why i disabled it @ukc. I ll see if IBM released a new bios. You certainly don't want to be shutting down power management completely, you need it to be working, and on that thing, you seem to need it working perfectly. Well i saw this to solve the sound to be crappy with pci power managed bug . Watching my T22 or T23 (whatever it is) running on ACPI was amazing -- the hw sensors were rather comprehensive, and you could watch the temp drop as you could feel the thing cooling off...it was obviously MUCH happier. Well , i ll give it a try , but it still would stay a PC :)) - http://cisc.free.fr/mapmusee.htm welcome to my personnal diner room where i kept various computers from an era where computer did not mean pc ... Cheers, Johan [demime 1.01d removed an attachment of type application/octet-stream which had a name of dmesg]
Re: Delete packages with dependencies
On Wed, Aug 19, 2009 at 02:02:07PM -0400, Luis Useche wrote: Do I have to do something else here? It seems like this discussion cooled down a bit. Is the patch in review? Is the patch been considered for inclusion? Are there any changes I can do to make more suitable for inclusion? I don't exactly how these things work and if I followed the regular path to submit the patch. Thanks, Luis You should direct your efforts in other directions. There are lots of things that are considered *not finished* in the current pkg tools, I'm slowly working on them, I have my list of stuff to do, and I won't look at your patches for now, because I have enough to tweak first, and getting this kind of functionality in right now is probably not a good idea. Ask again in six months, hopefully a lot of things will have settled down. I haven't ever had time to write the full picture of what's going on. Basically, I re-designed the pkg tools to accomplish certain goals, and offer certain guarantees. Perl means that I can do agile refactoring. My initial goal was to allow for updates, which no BSD supported at the time, to be as reliable as possible, and to be quite fast without building any extra information. The current design is surprisingly succesful at that. While working with updates, we discovered a few limitations, because some update patterns are just too weird to be true (dependency inversion, for instance), and because there's some stuff we did not predict (files moving from kdelibs to kdebase). So my current focus is to get these corner cases to work without having to cheat (which is what we currently do) and to get updates to feel faster (by merging the update discovery part with the actual update). This would also help people running -current, since one annoying feature of current is that you can't install new packages without updating your whole system first (the new update scheme would allow you to install a new package and check for updates on its dependencies only). The devil lies in the details: this is more complicated than it sound, so I'm refactoring more stuff to get rid of special cases so that it stays relatively bug-free (if you think about all the cases pkg_add currently handles, you'll notice that it's not THAT simple). So, yeah, I'm interested in feedback, but I can't be responsive on every part of it...
Re: MySQL unable to bind
On Sun, Aug 30, 2009 at 3:03 AM, Federico Giannicigiann...@neomedia.it wrote: We upgraded MySQL server from 4.1 to 5.0 in 3 servers. For various reasons (the most important is strict control of the versions) we always (since a lot of years) compiled the program from the sources. A few questions that would pop up are: 1. Are all three servers running OpenBSD? 2. If so, are they running the same version of OpenBSD? 3. When you say you compiled the program from sources do you mean you are using ports tree or do you really mean sources, as in download the sources manually, untar it and configure, make, etc? No problem at all with 2 of the servers, but we are unable to start the new server on the third PC, it always says that the address is already in use! Here is the exact error: 090830 10:08:16 [ERROR] Can't start server: Bind on TCP/IP port: Address already in use 090830 10:08:16 [ERROR] Do you already have another mysqld server running on port: 3306 ? netstat(1) and fstat(1) would be handy to see if any process is in fact using that port and which process it is. Though, the fact you are able to start the server as root user suggests there is something else wrong and this error message is erroneous. --patrick Obviously there is no other mysql server running! We stop it before installing the new version. Moreover, if we reinstall the 4.1 version it starts without any problem. I tried it tens of times! We use the exactly the same compile options on the three servers. Almost anything is the same: inetd.conf, login.conf, master.passwd mysql line, group, filesystem permissions on the logs, run and data directory... Today I have done some new experiments and found that the new server starts correctly if we let it run as root (user = root)! But obviously I don't want to let it run as root... So, the question is: what could make a normal (non root) user be unable to bind to a port that root user can bind? As the same server works on the other two PCs, there must be some subtle difference in this PC, but I really cannot find what! Anybody have any suggestion? Thanks.
