Hi,
Sometimes when I start CQRLOG 1.8.1 (Fedora) on older machines I get an error that cqrlog cannot connect to the database. Usually the next time I start it, it works just fine. I belive that there might be some timing issue on getting the mysqld started and then establishing the connection to it. Well, this did not bother me that much earlier but now I compiled cqrlog 1.8.1 for Fedora 20/ARM and this platform is slower than Intel/AMD. I keep getting this error pretty much every time I start cqrlog. It seems to start mysqld correctly but this probably takes too long time. I can connect to a remote database just fine (since it was running mysqld already) and cqrlog for that works OK. I compared the installed packages, running processes etc. to my intel Fedora 20 system and everything seems identical.
So, is it possible to increase the wait time between spawning the mysqld daemon and connecting to it?
Jussi Eloranta (AA6KJ)
ps. The standard Fedora src rpm for 1.8.1 compiles just fine on ARM. I had to remove the arm exclusion from the spec file to compile it. I will try to contact the Fedora people to let them know that it can now be compiled on ARM as well. This way we will get it into the standard ARM distro.
Sun, 2014-10-19 02:35
#1
mysqld startup
Hi,
yes, I can increase the wait time but it's not good solution, I have to find out something better. I already increased it and it seems, it didn't help.
Yes, CQRLOG works also on ARM, Ubuntu creates this packages for every Ubuntu release. I think, Debian build system also builds CQRLOG for ARM and even for PowerPC but I have no idea if it works or not, I don't have any ARM or PowerPC machine at home.
73 Petr
--
http://HamQTH.com/ok2cqr
https://ok2cqr.com
Hi Petr,
I haven't looked at the code but couple of things come to my mind:
1) A loop (with delay between iterations) that will try to establish the connection. This would be done several rounds before giving up. Those with fast server startup would get the connection immediately and would not be slowed down by this. One could perhaps add the delay and # of tries as parameters somewhere that users can change.
2) If the connection fails, put up a dialog that gives two choices: 1) try again or 2) quit. With this the user can decide how long to wait.
Jussi
ps. I just amended the update cqrlog request for fedora so that they would also to include the arm version (since it compiles just fine). Faster ARM boxes would not suffer from this startup issue. On intel it is part of the fedora distribution.
Hi,
a few minutes ago I pushed changes to CQRLOG repository on Github. Now CQRLOG is trying to connect to a database in a loop and there is 1 second delay between it. After 5 tries, cqrlog will show error message. Could you try it, please? Thank you!
73 Petr, OK2CQR
--
http://HamQTH.com/ok2cqr
https://ok2cqr.com