LHC@home

We are fully aware that our stats aren't working but there is a technical problem preventing us restarting them and without expert sysadmin effort we decided to leave it for now.

Ben Segal / LHC@home

---

Thanks for the confirmation / update. It's greatly appreciated!


William

---

I share the feeling of frustration. But we are not going to get good service from the incoming people if the first thing they see is a lot of complaints about how long they took to get here, and about stuff that happened before they were in post.

I'm sorry, but that irked me somewhat. It is worth remembering that they are getting a service from us, not the other way round. The incoming people need to realise that they are inheriting an existing entity, "warts an' all", they are not a "clean slate" startup operation.

To comment on the question in the did you all get work thread, the absence of XML stats makes the project look dead to those teams/consolidators that display them. As such, many team members will have simply detached or have LHC set to "No New Work".

Of course, some of the regulars will come back here when it gets going again, but in the hey day of LHC, it was competing with 3-4 other BOINC projects, now there are 30. It will be competing with communicative well run projects.

---

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.

We are fully aware that our stats aren't working but there is a technical problem preventing us restarting them and without expert sysadmin effort we decided to leave it for now.

Ben Segal / LHC@home

Thanks, I hope the resolution for this will come soon! I agree with the previous post about this issue - LHC is not an attractive project for new users to join because from third-party sites it appears to be totally inactive. This problem is compounded because oftentimes there is no work.

---

Idea: the huge amount of hosts may be the reason why the stats aren't working. /stats/host.xml would be VERY BIG. So they would have to solve that problem first.

---

Idea: the huge amount of hosts may be the reason why the stats aren't working. /stats/host.xml would be VERY BIG. So they would have to solve that problem first.

That might be part of it.
However the stats update failure seems to me to just part of a lack of communication that is very dissapointing.

Almost the last bit of news on the front LHC home page says "We will make every effort to keep you informed as this transfer proceeds.". And that's it, since 29th September! The new owners haven't even bothered to say "Hello. Sorry about this. We're trying to sort it."

As has been pointed out we are giving the project our services for free. In return we jus t expect a modicum of courtesy (tell us what's happening] and recognition of our efforts (exported stats please).

At present there are still more of us around than there is work to be done. I imagine that the surfeit of crunchers is getting less by the day.

---

At present there are still more of us around than there is work to be done. I imagine that the surfeit of crunchers is getting less by the day.

If there are 1000 workunits and 1000 users, it's possible that 10 users get 100 WUs each, and the rest nothing, instead of 1WU per user. Work doesn't get done faster like that... Like I said a dozen times (:P) a cache limit on LHC would be great.

---

Idea: the huge amount of hosts may be the reason why the stats aren't working. /stats/host.xml would be VERY BIG. So they would have to solve that problem first.

and they may or may not even know that is the issue, they'd just see the /stats directory filling the disk, and switch it off again till the proper admins arrive.

Definitely something worth a look.

R~~

It appears that stats are being exported from LHC.
Looks like the stats for individual teams runs o.k. as one of our team members has his own stats site that is updating with no issues at all.
But as river has pointed out exporting stats for so many hosts is looking like the cause of the problem.

---

Phanteks Luxe 2, AMD 3700x, MSI 2070 super X trio, 32gb Corsiar vengance RGB pro, watercooled.

your team member might be updating his stats manually by going to this site and copying the stats off the website, and not from the xml export. Looking in the stats directory, its still empty there.

---

lol twice the size of seti's lol seti's is 1.6Gb unzipped

If thats the case I bet Zain (our team Captain) is glad, the seti takes the server over 3 hours to process (when it works)

---

But as river has pointed out exporting stats for so many hosts is looking like the cause of the problem.

The praise is due to PovAddict - I just passed along (and agreed with) his diagnosis from another thread.

R~~

your team member might be updating his stats manually by going to this site and copying the stats off the website, and not from the xml export. Looking in the stats directory, its still empty there.

Or may be doing it by having a script that automatically goes to this website and extracts the stats. It is fine to do this for one or two users, or for the total stats for a team, but please don't try to do this automatically for all the individual members of a whole team - your script will hold up the db for a while, and/or provoke the "too many connections" issue.

It is a very tempting workaround for those of us who have the skills, but please resist the temptation.

River~~

...
However the stats update failure seems to me to just part of a lack of communication that is very dissapointing.

Almost the last bit of news on the front LHC home page says "We will make every effort to keep you informed as this transfer proceeds.". And that's it, since 29th September! The new owners haven't even bothered to say "Hello. Sorry about this. We're trying to sort it."
...

After our recent discussions with our Queen Mary College colleagues who will be taking over responsibility for LHC@home system administration, they estimate that it will be some time in January before the new server will be set up and running in London. Before then they will try to help out with sysadmin tasks remotely on the curent server. So there are no "new owners" yet, sorry.

Ben Segal / LHC@home

---

thx for the update

---

After our recent discussions with our Queen Mary College colleagues who will be taking over responsibility for LHC@home system administration, they estimate that it will be some time in January before the new server will be set up and running in London. Before then they will try to help out with sysadmin tasks remotely on the curent server. So there are no "new owners" yet, sorry.

Ben Segal / LHC@home

Ah, so we are still in limbo. But at least now we've an idea how long we might be here. This is helpful information and I thank you for it.
I still think it would be good if some sort of notice to this effect could be posted on the LHC homepage. Otherwise to the casual visitor it just looks as if everyone abandoned ship on 29th September. Possibly not everyone gets this far into the message boards?

---

After our recent discussions with our Queen Mary College colleagues who will be taking over responsibility for LHC@home system administration, they estimate that it will be some time in January before the new server will be set up and running in London. Before then they will try to help out with sysadmin tasks remotely on the curent server. So there are no "new owners" yet, sorry.

Ben Segal / LHC@home

...

I still think it would be good if some sort of notice to this effect could be posted on the LHC homepage. Otherwise to the casual visitor it just looks as if everyone abandoned ship on 29th September. Possibly not everyone gets this far into the message boards?

Ben,

I agree strongly with this point about the homepage.

A note on the main page to the effect that the server is expected to move in January, and that there is likely to be limited support till then would really help.

Then, whatever support is successfully achived by remote control will look like a bonus, rather than looking half-hearted.

At the same time, could you (or they) put up a note reminding folk that this project currently has work only rarely.

These two points on the homepage would help a lot.

Best regards,
River~~

Well said River~~.

Or may be doing it by having a script that automatically goes to this website and extracts the stats. It is fine to do this for one or two users, or for the total stats for a team, but please don't try to do this automatically for all the individual members of a whole team - your script will hold up the db for a while, and/or provoke the "too many connections" issue.

It is a very tempting workaround for those of us who have the skills, but please resist the temptation.

This is a somewhat contentious issue. Why does a single screen scraper running 3-4 times a day have a greater impact on the server then 100 team members doing the same thing 1-2 a day? The opposite is true I'd have thought.

---

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.

Why does a single screen scraper running 3-4 times a day have a greater impact on the server then 100 team members doing the same thing 1-2 a day? The opposite is true I'd have thought.

Only if 100 team members actually did it. The chances are that 99 of them would not bother. I guess when I said about an individual doing it, I meant that they wrote their own script and did not pass it on - that acts as a huge filter on who can be bothered to get the stats.

I agree, if a team writes a script and passes it round that is probably a worse effect than if a central team bot does the job.

hmmm.

If anyone is going to ignore my previous request and write a team bot to retireve the stats for all the members, the server friendly way to do it would be to take only the stats that are available on the team listings, because that way you get several at once. And please put in a 5min or so delay between each web fetch in your script, so that your query does not generate multiple db accesses close together.

But even better, please don't do it at all.

R~~

This is off topic now, but is a least staying civilised, so I'll pursue it.

Before I joined a team with team stats, I used to go to each project, (8-9 projects), 1-2 times a day to see what was happening. Most of the people I talked to did the same. It was a nuisence having to visit all the sites, but it IS what we were doing.

Now, I rarely bother looking at the project stats pages. Why? Because I can go to my teams stats page and get all of the stats for all of my projects in one place.

I now am the principle software developer for my team. I can use Ethereal to monitor my LAN traffic and can absolutely guarantee that the http GET requests the screen scraper sends are identical to what would be sent by an end user clicking on the relevent links on the project site. The screen scraper is like a single user that looks at their stats 3-4 times a day.

On the other hand, we now have an entire team of people who never look at the project stats anymore. This, to me, is clearly beneficial to the project, not derogatory.

When debated on S@H side, this degenerates quickly into a "you are the scum of the Earth" type bandwagon, but nobody has ever been able to rationally tell me what is wrong with this argument. Instead. all the reasonable people disappear from the thread, and the flame mongers have a day out.

What then happens is "screen scrapers are bad" gets ingrained into the "lore" of BOINC, and are universally decried as bad without any factual basis.

---

Wave upon wave of demented avengers march cheerfully out of obscurity into the dream.