Mika's down? Hey Mika...WAKE up and hit the ole reset button....:-)

[Engineer]

TeAm RR status...

When I posted this..Mika's in-buffer was at zero! As was BPhantoms and many of the others were on their way down....30k's out of 50K blocks...???

Just a heads up!

 

[TwoFace]

/me crosses fingers and hopes this will pass soon!

:|

Thanks for the headsup tho' Engineer!

With love and respect your fellow TA member

Two-Face

 

[jinsonxu]

Damn, i only have 3k in my buff-in.

 

[Engineer]

Update:

Apparantely Mika's pproxy is not giving/receiving blocks from D.net and the other RR pproxies apparantely feed off Mika's? (correct?)..

I tried to buffer from D.net directly and had no trouble from it?

So if anyone needs blocks, set the client to d.net and buffer them and then set back to team RR to dump!

Good Luck all!

 

[TwoFace]

Engineer: The Baby Bovines feed off Mika's that's correct

Something wrong with Mika's then I guess I hope whoever has the opportunity to remote control it or whatever it is they can do do so soon!

With love and respect your fellow TA member

Two-Face

 

[Dougal]

If anyone needs a quick fix I've got 30K still in my pproxy here - narnia.brother.co.uk.

I flush directly to Mika's so the stats should still be easy for all you statsaholics to trace.

 

[Jator]

I've got about 25,000 right now, though I am on the baby bovine rotation, so that will only last about a day.

/me kicks Mika to wake up.

 

[LeBlatt]

/me mails Mika about it in case someone else didnt yet (unlikely)

 

[alpha7x]

That would explain why Mika's proxy shows 4799 blocks received from me on Oct 25, and Dnet shows 992 received. Hope Mika's proxy gets connected soon...

 

[bphantom]

Morning guys,.. I just restarted the proxy. Looks like our upstream proxy is out of blocks. I switched it and restarted. Mika's is getting hammered pretty hard right now with requests. Wow!

P.S. My pproxy is also refilling as well.

Brad..

 

[LeBlatt]

Phew

 

[office boy]

Yikes that was scary.... I woke up and was cracking all randoms...

 

[JHutch]

This is a good example of why ALL my cows keep at least 24 hours worth of feed in their buffers. Keeps 'em cracking through these little "hiccups."

JHutch

 

[RaySun2Be]

I too keep 1 days worth in my proxy in buffer just for these occasions and some reserve in the cow's local in-buffer. Especially now you can set the newer clients to flush often even with large in-buffers.

One can either ingore Murphy and his laws, or you can do what I've learned (the hard way :Q), and that is to accept and embrace Murphy as a twin brother, and expect and plan like heck for his unannounced visits!



And even with that, I forget and he pops in for a surprise visit! :Q

 

[Crow]

I keep 3 days to cover problems over the weekend at the office

 

[LeBlatt]

Just a reminder : having a large in-buff in your cows wont make sure you always have a reserve. Even with an in-buff of 1 week worth, it may happen to be empty just at the beginning of an outage.

The best way to prevent that is to use a pproxy and set its minkeysready to the value that you want to always have at hand. Mine i set to min=12K and max=15K so whenever it gets under 12K it attempts to fetch 3K or so.



<< you can set the newer clients to flush often even with large in-buffers >>


How do you do that ? (sorry for the newbie question )

 

[dennilfloss]

The dialup Borg disk updates my buffers right after finishing a packet while connected, so if the outage occurs then, I'll still have ~1950+ WU to crunch. If it happens during one of my 8 hours sleep disconnect, I'll still have enough to crunch fo nearly 3 1/2 days.

 

[SSP]

Hehe... I kept 2000 blocks on my buff in. I fetch and flush directly to Jator's proxy, so I was saved (hehe.. next thing his proxy runs out of blocks as well... )

 

[Dougal]

LeBlatt, under Buffer options you have a selection called &quot;Additional buffer level checking&quot;. Set that to 3 and it will keep everything running in and out smoothly.

<edit> spelling was bobbins <edit>

 

[RaySun2Be]

LeBlatt,

Here is the info on Additional Buffer Level Settings from the -config option:


Additional buffer-level checking:

The following options are extensions to normal threshold management and are
not usually necessary:
0) no additional buffer-level checking. (default)
1) ensure that there is always work available.
2) ensure that all completed work is kept flushed.
3) both 1) and 2). (implied if 'Dialup detection options' are enabled)
4) update on per-project buffer exhaustion.

Options 1, 2 and 3 will cause the client to frequently check buffers levels.
(Frequency/interval is determined by the 'Buffer-level check interval' option)
You might want to use them if you have a single computer with a network
connection &quot;feeding&quot; other clients via a common set of buffers, or if you
want to ensure that completed work is flushed immediately.

Option 4 is a hint to the client to work on a single project as long as
possible (updating per-project buffers individually), rather than loop through
all active/enabled projects (one combined update per pass).

Here is the .INI parm:

[buffers]
frequent-threshold-checks=3

 

[jinsonxu]

So if i set it to 3
and dialup watcher is active, does that mean that the client will fetch and flush as soon as there are completed work units and a connection is detected?

 

[Engineer]

I've mentioned this before, but I run my three home PC's in a REMOTE Buffer Round Robin fashion. I keep 2K in each 400 MHz PC and 10K or so in the 1GHz PC . I then set each PC's keyserver to the teams RR pproxy. Then, I point PC1 to PC2 using Remote Buffer directory...PC2 to PC3 using the Remote Buffer Directory...and finally PC3 to PC1 using the Remote Buffer directory. All the blocks would have to be completely exhausted from all the PC's before any one would start on Randoms....

Maybe I'll start a pproxy soon....hmmmm?

It's now a GREAT day all!

Thanks Brad for the RESET!

 

[bphantom]

LeBlatt, I used to have a nice 3-4 day reserve. Well this problem will cause me to write a little script (today) which will watch the logs and let me know of any problems (page me, blink lights, who knows ). Since everyone likes stats: I just scanned through the logs and Mika's pproxy famine started a few minutes before 00:00 UTC yesterday evening and ran out three hours later 03:00 UTC. My pproxy was doing good at (25K in) 06:45 UTC, but got quite nicely hammered and ran out at 06:54 UTC. :Q I don't mind though, since at least one herd was well feed.

...Of course I just had to turn off my buffercheck script a couple days ago for maintenence. That would have restored my buffer back to normal levels. D'oh!

Brad..

 

[imported_CyberWire]

now maybe people will see why I buffer so much on my RR pproxy.. for times like this..

 

[ViRGE]

Cool, the RR setup passed it's first real-life test. Now that we now that the BBs can keep it fed until morning, and Mika doesn't get hammered as hard, we need to point more people to using the BBs if possible. I'll take a look at the final copy of the pproxy FAQ as soon as I can. In the meantime, thanks for saving our tanned hides BB ops.