You are not logged in.
Hi,
I put hg.slitaz.org / cook.slitaz.org to mitigate suspicious flood.
Infact many requests on cook.slitaz.org to fetch files/receipt/packages and flood requests on hg.slitaz.org
Need to investigate exactly happen, but to ensure forum and tank availability i put on 503 those vhosts.
Sorry for inconvenience.
Offline
Hi,
Sorry for maintenance, i test locally switch from apache cgi (spawn many python instances for requests) to wsgi mode.
Local tests seem good, i add apache-mod-wsgi and adjust vhost for hg on pangolin.
Need to monitor if CPU load increase but think not, infact load due of many python instances.
Infact when i put vhost hg maintenance and reload apache, load decrease and not came back.
For suspect flood, many windows hosts, we can't directly ban ip ranges, we have already limit rate by requests / requests_ip.
This flood impact host due of huge spawn python for hgwebdir cgi process.
Offline
Hi,
After more tests and read docs, threads, see that wsgi process need to be control for limit processes / threads.
I reenable hg vhost, ensure cgi module not loaded only wsgi, and limit threads / processes, now monitor to see that more efficient.
Offline
[ Generated in 0.024 seconds, 7 queries executed - Memory usage: 1.53 MiB (Peak: 1.77 MiB) ]