This instruction is for versions older than 1.0b.
Look the Configuration doc
1. Please, check if you have installed db. library from SleeppyCat.
2. If you have Linux, please, be sure you have got the right threads. All those which are not LinuxThreads are wrong ones. If you have got glibc2 be sure the version is fresh enough. During the work wrong threads unfold themselves by sudden transfer breaks. Wrong glibc – by the impossibility to serve more than 1024 requests. These problems are solved by installing the right libraries. For standard work under Linux a glibc version not lower than 1.1pl1 is required.
3. gzip -d < oops.tar.gz | tar xvf –
4. cd oops
The script was tested on Solaris (gcc and WorkShop cc), Linux (2.0.36+), FreeBSD (2.2.7+)
5. make make install cd /usr/local/oops
6. Edit oops.cfg to take into consideration the local conditions. Improve obligatorily the dns-servers addresses and names of files-storages of objects and of directories, where the stored URL database will be located. Introduce your networks to the group-s descriptions. All is documentary in the oops.cfg enclosed.
7. oops -c oops.cfg -z this command will prepare (format) files-storages for use.
8. oops -c oops.cfg And now you can check if it works:
Notes about some operational systems:
Linux – you’ll need the fresh glibc version (minimum glibc-2.0.7-xx where xx >=19) – in older glibc versions there were some problems with threads. Symptoms of threads’ incorrect operation are suddenly closed connections.
FreeBSD – I observed panic on 2.2.8 at stress load. As it was clarified later (thanks to V.Silyaev) it was a TCP_NODELAY problem. Currently this option under FreeBSD is locked.