> On 01 Jun 2015, at 10:57, Christian Mack <[email protected]> > wrote: > > Hello > > Am 2015-05-29 um 15:34 schrieb [email protected]: >> Thanks again for replying Christian, >> >> Well about this Error: >> >> May 29 10:08:12 sogod [16521]: <0x0x7f02b7276e90[SOGoCache]> an error >> occurred >> when caching value for key '[email protected]+attributes': "SERVER HAS >> FAILED AND IS DISABLED UNTIL TIMED RETRY" >> >> I read that it can be caused, because some parameters to have them not so >> well >> adjusted: >> >> for example: >> >> WOWorkersCount = 10; >> // Memcached >> SOGoMemcachedHost = 127.0.0.1; >> >> >> But actually, it seems that this part I have well configured; I am thinking >> more that this can be caused because I am using one of the minimum >> digitalOcean VM (512Mb Ram + 768Mb swap)... What I am not sure is about the >> consequences of this error. >> > > Did you actually start memcached? > If not, this is your error. > > Consequences will be: > - SOGo will be really really really slow > - Your DB accesses will increase > > > You do not want it to swap at all! > Access to swap is 100 to 1000 times slower than to RAM. > Because of that swap is only useful to prevent data loss, when RAM is > overloaded. > Get at least 4 GiB of RAM. >
I believe swap is used in this case since it is *cheap* replacement of RAM in cloud environment. 4GB RAM on DigitalOcean will cost considerably more than 512MB … I was successfully running SOGo with 512MB RAM + 512MB swap for single human user (+ 1 machine user only doing email). There were 3 concurrent connections from 3 devices for this user, was not using EAS though. Migrated to Amazon EC2 where minimum instance is 1GB RAM since then … so no current experience with 512MB RAM. Best Regards Martin.
smime.p7s
Description: S/MIME cryptographic signature
