Planning & Coordination for 2,2330L - Page 33

EdH · 2019-07-26, 13:16

Hey Seth,

The page looks great, especially the new additions, but I have one very, very, very minor thought and a personal request, if it's not too difficult:

Code:

This is iteration 6, Let me know if the thread if you want anything

Should the highlighted "if" be "in?"

My desktop is fine, but my tablet has trouble with the Individual Client Stats page. Could you add a link on the Hosts Stats page such that I can go directly to the first occurrence of my clients on the Individual page? Scrolling keeps invoking long hangs that keep telling me my browser has stopped working.

No real big issue, if it's troublesome.

Thanks for all the work,
Ed (math<X>)

SethTro · 2019-07-26, 20:44

Quote:

Originally Posted by EdH

Hey Seth,

My desktop is fine, but my tablet has trouble with the Individual Client Stats page. Could you add a link on the Hosts Stats page such that I can go directly to the first occurrence of my clients on the Individual page? Scrolling keeps invoking long hangs that keep telling me my browser has stopped working.

No real big issue, if it's troublesome.

Thanks for all the work,
Ed (math<X>)

I made a rather large change. You can now click on the host in the first tab and it *should* expand to show all the clients in that group (e.g. verbis<X> shows verbis1, verbis2...) . Let me know if this works and if you like it. I wrote the code really fast and am not super proud but if other people like it I'll keep it.

SethTro · 2019-07-26, 21:08

@VBCurtis

I wrote up a long first of possible features for Thomé I'm curious which would be useful for you (that I should maybe work on) and what else you might like.

Missing features at this time:
* List of outstanding WUs

* maybe counts of each type? count of failed, ... what do you look for here?

* sorting in the table (pretty simple with bootstrap but not done)
* Inspection of failed / bad stuff: number of bad WU, client "badness", ...

* Our client has failed a number of times, it might be worth exposing the last X lines of the log if it was easy to remove PII (IP addresses, client names, ...)

* What sorts of things do you look for here?

* Inspection of successful workunits
* rate of timedout tasks.

pinhodecarlos · 2019-07-26, 22:22

Can we all go full throttle for the next couple of weeks, vebis, seven, birch...?

EdH · 2019-07-27, 00:02

Quote:

Originally Posted by SethTro

I made a rather large change. You can now click on the host in the first tab and it *should* expand to show all the clients in that group (e.g. verbis<X> shows verbis1, verbis2...) . Let me know if this works and if you like it. I wrote the code really fast and am not super proud but if other people like it I'll keep it.

This is EXCELLENT! Much more than I asked for. It works great with my tablet with no delays. If you must do something else, I understand.

Thanks,

Ed

VBCurtis · 2019-07-27, 10:43

Quote:

Originally Posted by pinhodecarlos

Can we all go full throttle for the next couple of weeks, vebis, seven, birch...?

Carlos-
The database seems to crash multiple times per week with the current progress rate. It seems cado is not ready, at least with the default database package, for everyone to fire away. Also, I am not currently available to babysit and reset the server; user EdH has kicked it twice already this week, but this isnt his project and outages are likely to go unfixed for longer while I am away from my desktop.
So, I think we can expect no more than 50 million relations a day; rates higher than that appear to cause failures. I don't mind being wrong about this, but nobody should be surprised at an outage lasting 12 hours next time.
Besides, we're already on pace to finish within 100 days from the start of the project, quite a fast pace. It would be nice to get a little extra firepower at the end, when relations are roughly twice as hard to find!

lukerichards · 2019-07-27, 12:48

Quote:

Originally Posted by VBCurtis

Carlos-
The database seems to crash multiple times per week with the current progress rate. It seems cado is not ready, at least with the default database package, for everyone to fire away. Also, I am not currently available to babysit and reset the server; user EdH has kicked it twice already this week, but this isnt his project and outages are likely to go unfixed for longer while I am away from my desktop.
So, I think we can expect no more than 50 million relations a day; rates higher than that appear to cause failures. I don't mind being wrong about this, but nobody should be surprised at an outage lasting 12 hours next time.
Besides, we're already on pace to finish within 100 days from the start of the project, quite a fast pace. It would be nice to get a little extra firepower at the end, when relations are roughly twice as hard to find!

Useful info - I was considering firing up the 12 cores I have at home for a few weeks, but I think I'll save myself the electricity bill in this case!

SethTro · 2019-07-27, 19:29

Quote:

Originally Posted by VBCurtis

Carlos-
The database seems to crash multiple times per week with the current progress rate. It seems cado is not ready, at least with the default database package, for everyone to fire away. Also, I am not currently available to babysit and reset the server; user EdH has kicked it twice already this week, but this isnt his project and outages are likely to go unfixed for longer while I am away from my desktop.
So, I think we can expect no more than 50 million relations a day; rates higher than that appear to cause failures. I don't mind being wrong about this, but nobody should be surprised at an outage lasting 12 hours next time.
Besides, we're already on pace to finish within 100 days from the start of the project, quite a fast pace. It would be nice to get a little extra firepower at the end, when relations are roughly twice as hard to find!

Do you mind if I download some of the work unit data and try to investigate the source of this error? My plan is to create a replay tool and see if I can investigate where the server hangs.

VBCurtis · 2019-07-28, 02:24

Quote:

Originally Posted by SethTro

Do you mind if I download some of the work unit data and try to investigate the source of this error? My plan is to create a replay tool and see if I can investigate where the server hangs.

Be my guest, Seth. I would like to continue to run these team cado projects, so reducing instabilities is a worthy pursuit.

pinhodecarlos · 2019-07-28, 08:07

Noted Curtis.

Seth, is it possible to add the number of threads on the active clients (cumulative)?

SethTro · 2019-07-28, 18:05

Quote:

Originally Posted by pinhodecarlos

Noted Curtis.

Seth, is it possible to add the number of threads on the active clients (cumulative)?

I'm not super interested in the threads because of hyper-threading and the wide range of Rel(s)/CPU-sec.

Technically you can make guesses from the ratio of cpu-seconds / time from last workunit or from looking at -t in command (if it's overridden). If you're interested and want to figure it out I'm happy to look at a pull request.

Code:

select DISTINCT substr(resultclient, 0, INSTR(resultclient, '.')) , substr(wu, INSTR(wu, '-t ') , 5) from workunits order by 1,2;

Sadly this return -t 4 for all work units.

2019-07-26, 13:16	#353
EdH "Ed Hall" Dec 2009 Adirondack Mtns 110111010100₂ Posts	Hey Seth, The page looks great, especially the new additions, but I have one very, very, very minor thought and a personal request, if it's not too difficult: Code: This is iteration 6, Let me know if the thread if you want anything Should the highlighted "if" be "in?" My desktop is fine, but my tablet has trouble with the Individual Client Stats page. Could you add a link on the Hosts Stats page such that I can go directly to the first occurrence of my clients on the Individual page? Scrolling keeps invoking long hangs that keep telling me my browser has stopped working. No real big issue, if it's troublesome. Thanks for all the work, Ed (math<X>) Last fiddled with by EdH on 2019-07-26 at 13:17 Reason: Because I wanted to...

2019-07-26, 21:08	#355
SethTro "Seth" Apr 2019 5×47 Posts	@VBCurtis I wrote up a long first of possible features for Thomé I'm curious which would be useful for you (that I should maybe work on) and what else you might like. Missing features at this time: * List of outstanding WUs * maybe counts of each type? count of failed, ... what do you look for here? * sorting in the table (pretty simple with bootstrap but not done) * Inspection of failed / bad stuff: number of bad WU, client "badness", ... * Our client has failed a number of times, it might be worth exposing the last X lines of the log if it was easy to remove PII (IP addresses, client names, ...) * What sorts of things do you look for here? * Inspection of successful workunits * rate of timedout tasks.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Coordination thread for redoing P-1 factoring	ixfd64	Lone Mersenne Hunters	80	2021-01-19 17:17
big job planning	henryzz	Cunningham Tables	16	2010-08-07 05:08
Sieving reservations and coordination	gd_barnes	No Prime Left Behind	2	2008-02-16 03:28
Sieved files/sieving coordination	gd_barnes	Conjectures 'R Us	32	2008-01-22 03:09
Special Project Planning	wblipp	ElevenSmooth	2	2004-02-19 05:25

2019-07-26, 22:22	#356
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 37·131 Posts	Can we all go full throttle for the next couple of weeks, vebis, seven, birch...?

2019-07-28, 08:07	#362
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 37×131 Posts	Noted Curtis. Seth, is it possible to add the number of threads on the active clients (cumulative)?