LHC@home

It appears that LHC may be creating a new host entry, every time a host connects.

Current host numbers being allocated have jumped significantly in recent months.

Although I have only 5 physical hosts, I found I had many thousands of hosts entries under my profile, many created only a few minutes apart.

---

that happened to me also, but there is an easy fix
look in this thread for the solution

http://lhcathome.cern.ch/forum_thread.php?id=2273

It appears that LHC may be creating a new host entry, every time a host connects.

Current host numbers being allocated have jumped significantly in recent months.

Although I have only 5 physical hosts, I found I had many thousands of hosts entries under my profile, many created only a few minutes apart.

---

that happened to me also, but there is an easy fix
look in this thread for the solution

http://lhcathome.cern.ch/forum_thread.php?id=2273

click here to go there

R~~

PS watnou you may like to look at this posting to see how to make a link - if you reply the system will show you how I did it.

well
i think that people do know how to use 'copy and paste' but a clickable url for those who can't is indeed handy

that happened to me also, but there is an easy fix
look in this thread for the solution

http://lhcathome.cern.ch/forum_thread.php?id=2273

click here to go there

R~~

PS watnou you may like to look at this posting to see how to make a link - if you reply the system will show you how I did it.

---

I also have a hundred hosts in my list. In fact, I just got three workunits, and Your computers page shows three of my hundred hosts with one result each! What's going on here?
My last host ID is 2859945, but user IDs are in the factor of thousands. So we have around a thousand hosts per user average! This is probably making server load worse, hosts table must be HUGE.

---

PovAddict wrote:

My last host ID is 2859945, but user IDs are in the factor of thousands. So we have around a thousand hosts per user average!

2860130 :lol: slowing down now (I thought i'd fixxed that pc too)

Wish I was 'Average' :lol: long as these new pc dont show up on my power bill

---

long as these new pc dont show up on my power bill

Hahaha

---

Max host number I found: 2866622. They are going up faster than I thought. How I know? well, trial and error changing ID http://lhcathome.cern.ch/show_host_detail.php?hostid=here. Once I found the lowest ID where it says "Couldn't find computer", I refreshed and it existed. Started going up one ID at a time and found again couldn't find computer. Waited a minute, refreshed and started increasing ID again, there were more than 20 new hosts!

EDIT: 2866666 now. And I think it was my searching what caused "Too many database connections" a minute ago, sorry! :[

EDIT2: at 18:38, 2866750, this is insane.

EDIT3: at 18:45, 2866850, you can see the rate there

EDIT4: at 19:00, 2867090 hosts. There is a workaround here, but not all users will do it, so server is still going to die once hosts table fills the server disks.

---

Bump, so you notice edits. Editing doesn't cause thread to show unread sooo...

---

I just attached to LHC again, and got host ID 3415083 assigned. I stopped BOINC and set it on client_state.xml. But what scares me is the number, getting way too high! Can't they disable the scheduler while there is no work or something? Won't solve the problem but will avoid getting SO many hosts.

In fact I would prefer scheduler off until problem is solved.

---

This is still a problem -- tried attaching a laptop for a couple of days -- without checking xml file to ensure that it updated with a host id and guess what --- it didn't ;/ so had to merge almost 500 hosts into one so i could delete it as wife took the laptop away.

Hope this is something that gets looked at otherwise all new users/hosts are going to get LOTS of hosts listed -- like every time they connect.

Larry

---

I'd suggest blocking the scheduler from receiving connections while there is no work to send and no work in progress. That would reduce some of the host duplication (for those people who never look at forums to know the clientstate.xml fix). I think it's a high-ish priority to avoid even more hosts to be created, there are already too many, and I'm sure the admins will have problems later with the database if this grows worse.

---

Here's some fresh data: LHC has 5531003 host entries.

---

not everyone will notice that they have a zillion of hosts.

i noticed it myself a few days ago that since januari my hosts were multiplying.

.....for those people who never look at forums to know the clientstate.xml fix.....

---

I just checked my old posts and it has been 233 days since I posted detailed
instructions how to fix this from your client.

As soon as the new server is up and running the current release of boinc software
(v5.09) it will no longer be an issue.

Steve

---

98SE XP2500+ @ 2.1 GHz Boinc v5.8.8

Good to hear that is a server software issue as just had it happen on boincmgr 5.8.11 and was wondering. At least this time i remembered and only let it contact twice before editing the clientstate.xml file

Larry

---

... I think it's a high-ish priority to avoid even more hosts to be created, there are already too many, and I'm sure the admins will have problems later with the database if this grows worse.

I agree. Upgrading to the latest server software should fix this problem and possibly a few others. I would rate that as the number one priority after getting moved to the new server.

Getting rid of the old hosts should probably also be high on the priority list. That will probably have to be done before the stats can be exported. With as many hosts as have been generated and since many participants don't know to merge them, it would be nice to have an automated script that merges hosts.

---

BOINC WIKI

BOINCing since 2002/12/8

Updating the server may help. I have had this same problem on PrimeGrid & ExtremLab is reporting it in their news. They have seen it linked to an update of the Boimc client on the home machine.

This is how I got the problem, updating to Boinc 5.8.11. The symptom was a new CPID every day, a new PrimeGrid host entry every day and one machine connected to PrimeGrid as two seperate & independent computers. It also had PrimeGrid listed twice in the Boinc manager project list. It didn't affect any other project in my case.

If my experience is any indication I'd say this is likely a fairly common headache.

---