Message boards : Number crunching : Merging "unmergable" hosts
Message board moderation
Author | Message |
---|---|
Send message Joined: 17 Sep 04 Posts: 25 Credit: 196,284 RAC: 0 |
Hi everybody, I recently noticed that there were multiple hosts related to my CPDN account although I'm using one and only one computer for BOINC. This is because "Sometimes BOINC assigns separate identities to the same computer by mistake" as stated elsewhere on the Web site. There is also an option to merge such multiple hosts into one. I had 5 different hosts for CPDN, and I managed to merge them into two distinct hosts, 67836 and 88669, but I cannot merge these two although they are exactly identical. Such was the case for other projects, and I managed to merge all those multiple hosts into one; now I have one host for all other projects, two hosts for CPDN. This is a trivial matter, but I'm curious as to why this might happen, and any suggestions to merge the remaining two CPDN hosts are welcome. Regards, Ertugrul. |
Send message Joined: 26 Aug 04 Posts: 59 Credit: 438,133 RAC: 0 |
Looking at your computers list, I saw that the processor is called by two different names - this might be the problem. Have you overclocked the pc at any time? |
Send message Joined: 17 Sep 04 Posts: 25 Credit: 196,284 RAC: 0 |
Yes, that's quite right, the new one is "GenuineIntel Intel(R) Pentium(R) M processor 1400MHz" whereas the old one is "Intel(R) Pentium(R) M processor 1400MHz Pentium". Device Manager says it is "Intel(R) Pentium(R) M processor 1400MHz Pentium", too. I don't understand where that "Genuine..." came from, but that seems to be the most recent one. Most probably that's why I had more than one host in other projects, too. But I managed to merge them, whereas I cannot merge these two in CPDN. No, I haven't overclocked my notebook, nor did I have the slightest intention to do so... Any other suggestions? |
Send message Joined: 26 Aug 04 Posts: 59 Credit: 438,133 RAC: 0 |
I'm sorry, no :-( I've had the same problem on some projects and don't really know why the different naming happens. I've heard that it could happen if you overclock the system, hence my question. I hope someone else has an explanation. |
Send message Joined: 26 Aug 04 Posts: 100 Credit: 1,191,715 RAC: 0 |
Have you updated BOINC? I think I saw that one of the bugs fixed recently was correcting processor detection. ____________________________<br> <a href="http://www.boincforum.info/boinc/">boinc forum</a> and <a href="http://www.uk4cp.co.uk/">United Kindom</a> team, my climate change <a href="http://www.livejournal.com/users/mike_atkinson/">blog</a>. |
Send message Joined: 17 Sep 04 Posts: 25 Credit: 196,284 RAC: 0 |
I'm running 4.19 which I believe is the latest version as of today. Anyway, there is nothing I can do other than merging the hosts if an old buggy version has misidentified my processor, which I cannot do... I believe a few SQL statements will do the trick, so if an admin or "super user" is reading this, maybe... :-) |
Send message Joined: 17 Sep 04 Posts: 25 Credit: 196,284 RAC: 0 |
I think I found out why the project assigned different host IDs. Here is an excerpt from <a href="http://boinc.berkeley.edu/host_id.php">BOINC Web Site - Host Identification</a>: Not all systems have an application-readable globally unique IDs, so we use a different approach to host identification. When a host first contacts the scheduling server it is assigned a host ID. The server also maintains an RPC sequence number for each host. Both the host Id and the RPC sequence number are stored in the client's client_state.xml file. If the scheduling server receives an RPC with a sequence number less than the expected sequence number, it creates a new host record in the database. (This might happen, for example, if a user copies the client_state.xml file between hosts.) I occasionally take backups to avoid losing WUs in case CPDN fails completely. That happened a few times, and I had to resort to my backups which had RPC sequence numbers less than what BOINC expected, hence new host IDs. But I still have to figure out how to merge the remaining two hosts. Cheers, Ertugrul. |
Send message Joined: 5 Aug 04 Posts: 186 Credit: 1,612,182 RAC: 0 |
Carl Christensen effectively fixed a similar problem with Classic CPDN by allowing selective 'hiding' of hosts that folks no longer wanted to see in their stats - a similar function in BOINC's 'Your Account / My computers' page would be rather nice to have; I use that page to check on which computers are still running efficiently and <a href="http://climateapps2.oucs.ox.ac.uk/cpdnboinc/hosts_user.php?userid=156">it's gotten rather confused</a> with all the changes I've made recently - I only have 8 machines online at the moment but <i>18</i> are listed. :-( I already have quite a few 'unmergeable' machines (eg. host 61912 & host 61968 which are exactly the same machine except 61968 has SP2...) and doubtless I will end up with even more during the switch to BOINC v4.19 although 'Alison' (Host 42174) <i>has</i> transitioned her processor name change okay. :-) <a href="http://www.nmvs.dsl.pipex.com/"><img src="http://boinc.mundayweb.com/cpdn/stats.php?userID=6&team=off&trans=off"></a> <a href="http://www.nmvs.dsl.pipex.com/">Distributed Mania</a> |
Send message Joined: 4 Sep 04 Posts: 14 Credit: 468,276 RAC: 0 |
> Carl Christensen effectively fixed a similar problem with Classic CPDN by > allowing selective 'hiding' of hosts ... UK_NICK I believe this may work... Create a backup of CPND use browser to open "unmergable" http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=xxxx (your "lower ID host") use browser to open "current" http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=yyyyy (your "higher ID host") replace parameters values such as <p> (CPU type from higher ID host), <p> (Measured floating point speed from higher ID host), <p> (Measured integer speed from higher ID host), for example Microsoft Windows XP, for example Professional Edition, Service Pack 2, (05.01.2600.00), (from the "lower ID host"), (if different) in client_state.xml, I think I also changed it in the client_state backup file... Make corresponding changes in sched_request.xml save files and also make a copy of the (3) modified xml files in a different location... unplug network card spark up CPND if the bench mark run automatically - let it finish, then suspend and exit CPND and recopy 3 modified files to CPND directory spark up CPND connect network do a manual update just for good measure wait few minutes login and you should be able to merge your "unmergable" lower ID host with your current higher ID host Oh - just a reminder, I am doing this from memory - to be used at own risk:) |
Send message Joined: 17 Sep 04 Posts: 25 Credit: 196,284 RAC: 0 |
> > I believe this may work... > > Create a backup of CPND > use browser to open "unmergable" > http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=xxxx (your "lower ID host") > use browser to open "current" > http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=yyyyy (your "higher ID host") > replace parameters values such as > > (CPU type from higher ID host), > > (Measured floating point speed from higher ID host), > > (Measured integer speed from higher ID host), for example Microsoft Windows XP, > for example Professional Edition, Service Pack 2, (05.01.2600.00), (from the "lower ID host"), (if different) > in client_state.xml, I think I also changed it in the client_state backup file... I'm assuming that what you describe here is supposed to trick BOINC server into thinking that the old and the new hosts are the same by supplying the same values for a certain field. But it doesn't work for me, let me explain... In my case only CPU type is different (I'm not counting CPU and upload/download speeds, I think it is normal that they are different,) and I did as you advised, but the client changes the CPU type in the client_state.xml and the backup file to that of the current "higher ID" host automatically whenever the client starts up or a manual update is made. I think the difference in CPU type value is due to the fact that the new client versions identify processors more accurately. Any other suggestions or something else not from memory? Do you have any idea as to how the related PHP code decides that tho hosts are "unmergable"? Cheers, Ertugrul. |
Send message Joined: 17 Sep 04 Posts: 25 Credit: 196,284 RAC: 0 |
This one was the same as the previous one, I submitted the same post twice by mistake. Sorry... |
©2024 cpdn.org