Hi, I'd like to update this thread too by trying last 2.0-M2-SNAPSHOT where now it's asynchronous apart form Hazelcast.
Lvc@ ᐧ On 9 September 2014 11:24, <vogt....@gmail.com> wrote: > So here is my final report... > > We have scoured all the available docs and come to the conclusion that > OrientDB *does not* support 100% asynchronous replication. > > writeQuorum 1 is just as fast as writeQuorum 2 which leads me to believe > this flag has no effect in the sources. Alternatively we have also tried to > follow the WAN Replication features of Hazelcast at > http://hazelcast.org/docs/3.1/manual/html/ch12.html, however OrientDB > does not seem to do any replication in this mode. The docs aren't really > clear on the steps involved other than alledging asynchronous replication > is achieved by simply setting writeQuorum to 1. > > I don't know what other people do with OrientDB, but having our servers in > sync is a must have feature for us. We were looking at purchasing the > Enterprise Edition, though with the lack of feedback arising from this > issue decided against doing so. > > Perhaps in the future OrientDB will support the features we require. > > > > On Saturday, September 6, 2014 4:43:36 PM UTC+2, vogt...@gmail.com wrote: >> >> Hi Luca, >> >> I've just tested orientdb-community-1.7.9-20140906.091114-5 SNAPSHOT and >> unfortunately I see no improvement. I'm inclined to open an issue on >> github, but I have no idea if this issue is actually worked on or not. >> >> Please advise if there is anything else you need from my side. I would be >> glad to assist. On the other hand, if everything appears to be working as >> it should, I would appreciate if you could let me know so we can move on. >> >> Thanks for your understanding. >> >> Regards, >> - Ben >> >> On Friday, August 22, 2014 1:13:32 PM UTC+2, vogt...@gmail.com wrote: >>> >>> Hi Luca, >>> >>> As requested, dserver test (writeQuorum 1): >>> >>> orientdb {GratefulDeadConcerts}> jss for(i=0;i<10000;i++) >>> {db.command('insert into followed_by (out, in, weight) values (#9:1, #9:1, >>> 1)');} >>> [Started multi-line command. Type just 'end' to finish and execute] >>> orientdb {GratefulDeadConcerts}> end >>> >>> Server side script executed in *7278.426758* sec(s). Value returned is: >>> followed_by#11:17663{out:#9:1,in:#9:1,weight:1} v1 >>> >>> Results in avg of *0.72784*.. seconds per insert. >>> >>> In contrast, server: >>> >>> orientdb {GratefulDeadConcerts}> jss for(i=0;i<10000;i++) >>> {db.command('insert into followed_by (out, in, weight) values (#9:1, #9:1, >>> 1)');} >>> [Started multi-line command. Type just 'end' to finish and execute] >>> orientdb {GratefulDeadConcerts}> end >>> >>> >>> Server side script executed in *1.995000* sec(s). Value returned is: >>> followed_by#11:27663{out:#9:1,in:#9:1,weight:1} v1 >>> >>> Results in avg of *0.00019*.. seconds per insert. >>> >>> Regards, >>> - Ben >>> >>> On Thursday, August 21, 2014 11:19:07 AM UTC+2, Lvc@ wrote: >>>> >>>> Hi Ben, >>>> Sorry for delay. That is not the time of each insert. Please try to do >>>> the same in a loop of 10k times and get the average time per cycle. >>>> >>>> However I'm working on a separate branch to speed up distributed >>>> configuration. We have good numbers so far, but we need more days to merge >>>> with 1.7.9. >>>> >>>> Lvc@ >>>> ᐧ >>>> >>>> >>>> On 21 August 2014 10:04, <vogt...@gmail.com> wrote: >>>> >>>>> Dear Luca, >>>>> >>>>> I've sent you the log last Friday. I urgently need to move forward on >>>>> this. If nothing can be done to get OrientDB to work asynchronously, I >>>>> have >>>>> to look at alternatives. >>>>> >>>>> OrientDB appears interesting for numerous reasons though, which is why >>>>> I'd like to stick with it. >>>>> >>>>> Regards, >>>>> - Ben >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> On Thursday, August 14, 2014 11:01:42 PM UTC+2, Lvc@ wrote: >>>>> >>>>>> Hi Ben, >>>>>> setting writeQuorum = 1 in your scenario means that the server node >>>>>> where the client is connected should be the first to respond (for obvious >>>>>> reasons), so it's crazy it takes 1 second for 1 insert. >>>>>> >>>>>> To know more try to enable logging on the server where the client is >>>>>> connected. Edit config/orientdb-server-log.properties and put these >>>>>> lines >>>>>> >>>>>> .level = INFO >>>>>> com.orientechnologies.level = INFO >>>>>> com.orientechnologies.orient.server.distributed.level = FINE >>>>>> com.orientechnologies.orient.server.hazelcast.level = FINE >>>>>> com.orientechnologies.orient.server.journal.level = FINE >>>>>> >>>>>> At the restart you should have much more logging information, then >>>>>> send us the log for the insert. >>>>>> >>>>>> Lvc@ >>>>>> >>>>>> >>>>>> >>>>>> On 14 August 2014 20:38, <vogt...@gmail.com> wrote: >>>>>> >>>>>>> Dear colleagues, >>>>>>> >>>>>>> I am trying to figure out how to deploy a distributed orientdb >>>>>>> setup. I've followed the manuals, which explicitly state that >>>>>>> writeQuorum 1 >>>>>>> will result in 100% asynchronous mode, however I am faced with inserts >>>>>>> taking up to 1 second or more. Example: >>>>>>> >>>>>>> orientdb {GratefulDeadConcerts}> insert into followed_by (out, in, >>>>>>> weight) values (#9:1, #9:1, 1) >>>>>>> >>>>>>> Inserted record 'followed_by#11:7061{out:#9:1,in:#9:1,weight:1} >>>>>>> v1' in 0.953000 sec(s). >>>>>>> >>>>>>> When I don't run in distributed mode, I have: >>>>>>> >>>>>>> orientdb {GratefulDeadConcerts}> insert into followed_by (out, in, >>>>>>> weight) values (#9:1, #9:1, 1) >>>>>>> >>>>>>> Inserted record 'followed_by#11:7063{out:#9:1,in:#9:1,weight:1} >>>>>>> v1' in 0.002000 sec(s). >>>>>>> >>>>>>> writeQuorum 1 has no effect, I've tried several releases from the >>>>>>> 1.7 branch and even 2.0, all with the same result. It is important to >>>>>>> note >>>>>>> that the nodes are each located on a different continent. >>>>>>> OrientDB looks great and I would really like to use it. Any advice >>>>>>> on how to achieve asynchronous replication is greatly appreciated. >>>>>>> >>>>>>> Best regards, >>>>>>> >>>>>>> - Ben >>>>>>> >>>>>>> -- >>>>>>> >>>>>>> --- >>>>>>> You received this message because you are subscribed to the Google >>>>>>> Groups "OrientDB" group. >>>>>>> To unsubscribe from this group and stop receiving emails from it, >>>>>>> send an email to orient-databa...@googlegroups.com. >>>>>>> >>>>>>> For more options, visit https://groups.google.com/d/optout. >>>>>>> >>>>>> >>>>>> -- >>>>> >>>>> --- >>>>> You received this message because you are subscribed to the Google >>>>> Groups "OrientDB" group. >>>>> To unsubscribe from this group and stop receiving emails from it, send >>>>> an email to orient-databa...@googlegroups.com. >>>>> For more options, visit https://groups.google.com/d/optout. >>>>> >>>> >>>> -- > > --- > You received this message because you are subscribed to the Google Groups > "OrientDB" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to orient-database+unsubscr...@googlegroups.com. > For more options, visit https://groups.google.com/d/optout. > -- --- You received this message because you are subscribed to the Google Groups "OrientDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to orient-database+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
