Is somebody on Mars time?

---

You know it is for DA to fiddle and make people happy ;-)

---

And they're down, 5:40am Pacific.

& still down 10 hours later.

And they're down, 5:40am Pacific.

& still down 10 hours later.

Early to sleep and late to rise,
makes a project ....

---

Still down, are we going for a record?

Still down, are we going for a record?

Up!

---

How disappointing, we came back by 5:30 PST, so we were a bit short of 12 hrs.

Glad I didn't stay up to see the end

SETI is down hard:

This site can’t be reached
setiathome.berkeley.edu took too long to respond.

...Asteroids just goes down for days on end with messages like "Project is temporarily down for maintenance" whenever you connect, and nothing on the front page....

If the project is not stable after a 12 hour outage, then there is a fundamental problem somewhere. No other Boinc project needs to be off air for a whole day to do database and server backups.

Chris, There is a fix for that; there is a button labeled "Remove". If you press it I assure you you will never have any more issues with outrages.

---

... & you're missing the point. They are operating on a shoestring & the staff have a life to live.

Unlike major companies with big I.T departments, they are only a small team. Don't forget Amazon & "that" typo.

Einstein has redundancy with having their backup servers available in different parts of the world, so there is always one you can download from and upload to/report to. And when your normal server is down, BOINC will automatically switch to the next nearest one to try on the next contact.
Einstein also actively checks their data for the things they are looking for, and checked data is either discarded of or added to a database that isn't actively associated with the project.

Seti however has everything together in one place, both the primary as the secondary server. It also has several terabytes of data to backup, not an easy feat. When you're doing home backups of your system you know that backing up several hundreds of gigabytes can already take an hour or more, so what do you think several terabytes will set you back? And it's not just the backup, you'll have to check the integrity of that backup as well. Extra hours.

Seti may now have a good way of checking the data they accumulated since 1999 - with Nebula - but they still need to keep all that data together. And it is added to on a daily basis.
As for sizes of database and files and how long everything takes to check with Nebula, do start reading from https://setiathome.berkeley.edu/nebula_web/storage.php onwards.

---

Although that is certainly true, I don't think it answers Chris's point.

Both SETI and Einstein have similar server topologies: there are a pair of 'transactional' (BOINC) servers - primary and read-only replica - and a separate scientific repository for completed and validated scientific work.

Only the transactional servers are visible in the sense that Chris and other volunteers from the general public are asking about - why do (or don't) they need regular maintenance?

Matt Lebovsky answered that in SETI's technical news threads, many years ago - too long ago for me to be bothered to look up the exact reference. But in essence, he wrote that the actual backup of the BOINC server took only a small part of the maintenance interval: the majority of the time was spent compacting and re-indexing the transactional tables so that the database could function efficiently for the next operational cycle.

I suspect that the major difference between the two projects is the transaction rate, and hence the degree of database fragmentation, together with the sheer size of some of SETI's more slowly changing tables. Einstein is sending out and receiving ~1.3 million workunits per week: SETI is handling the same number per day (typically over 110,000 results per hour, these days). Re-indexing SETI's tables will have to include the vastly larger user, host and message board archives: Einstein has taken steps to reduce their data overheads (e.g. combining multiple data segments into fewer, larger, workunits, and de-coupling the message boards from the BOINC database framework).

But the biggest difference is in resourcing. Einstein (by my count) credit 14 contributors across their Wisconsin and Hanover sites, where SETI only credits six - and at least half of those are in supervisory, rather than operational, roles. If SETI had a similar staff-to-workload ratio as Einstein, I'm sure they could work out some optimisations similar to Einstein's, but for the time being they merely have the capacity to stop the wheels falling off the BOINC wagon - and although we can no longer ask Rom Walton for help with database optimisation, I suspect BOINC's data structure was never designed in the expectation of handling numbers on this scale. Remember how SETI had to make some emergency alterations a few years ago, when first task and then workunit IDs outgrew the 31-bit address space allowed?

In a nutshell then:

They are operating on a shoestring. Unlike major companies with big I.T departments, they are only a small team.

Amazing what can be conveyed in a short sentence or two.

In a nutshell then:

They are operating on a shoestring. Unlike major companies with big I.T departments, they are only a small team.

Amazing what can be conveyed in a short sentence or two.

In other words the 1st rule of engineering Keep It Simple Stupid ;-) , 2nd rule is if it don't work hit it with a hammer , if that hammer don't work get a f###ng bigger hammer ( that's what probably goes on during the outrages but we won't ever know ;-) )

But it also has to be remembered that Seti is a UCB funded on Campus project.

No it isn't. It is entirely donation funded and UCB skims the donations.

---

https://www.youtube.com/watch?v=Xf1S9U2ghrI

Nope, you are not quite worth my voice.

Commonly known in the trade as a Birmingham screwdriver!!

I know that like you do Chris but wern't that sure many others would know , perswading stick is another good name for wackin sticks ;-) ( i got a full set all the way from chipping hammer to sledge ;-) )