| Author |
Message |
|---|
 Matt LebofskyVolunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: 1 Mar 99
Posts: 1375
Credit: 74,079
RAC: 0
|
|
Regular backup outage today. Everything was normal except we bounced the replica database to change one buffer size setting and now nobody can connect to it - even to shut it down! Seems like we lost all our connection permission info somehow. From what we can tell it is still acting as a replica and making updates, but we can't access the data at all. We're stumped. Bob's looking into it.
On the plus side, we got all the pieces in place to move another function off kryten and onto bruno: file deletion. I just fired this off, and at first glance it seems faster. Time will tell if this is an improvement. Bruno is a faster machine in general, but kryten had a gigabit connection to the workunit file server, while due to lab infrastructure bruno can currently only have 100 Mbit. So we shall see. Hopefully queues will drain after we recover from the outage backlog.
Here's a fun one: Since the switchover to using Hurricane Electric as our main ISP I noticed lingering traffic on the campus router which served our Cogent link. We're talking as much as 1 Mbit/sec. Today while updating lab-wide DNS records I noticed shserver2 was still there. This was the DNS alias for our SETI@home classic data server. I removed this entry, and check out the dip in traffic:
So, well over a year since unplugging the classic data server, there are still enough SETI@home classic clients around the world trying to access a missing server to account for almost 1 Mbit/sec of traffic on UC Berkeley campus routers. Not sure how to exactly explain the shape of this graph (and why incoming = outgoing). The diurnal shape and hourly ridges look like scripts or cronjobs running on machines that haven't been checked in ages.
A lot of BOINC naysayers like to point out how many classic users "quit" last year after the big transition. But this graph adds some meat to my theory that a large chunk of the SETI@home classic users actually left the project many ages ago, and their old clients simply continued to run unattended. Mind you, this is 1 MBit/sec of traffic without actual workunit data being sent - just SYNs, basically. I think. Somebody break out a calculator and determine how many SETI@home classic clients this represents.
- Matt
____________
-- BOINC/SETI@home network/web/science/development person
-- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude |
|
|
|
|
Here's a fun one: Since the switchover to using Hurricane Electric as our main ISP I noticed lingering traffic on the campus router which served our Cogent link. We're talking as much as 1 Mbit/sec. Today while updating lab-wide DNS records I noticed shserver2 was still there. This was the DNS alias for our SETI@home classic data server. I removed this entry, and check out the dip in traffic:
Matt,
There is a flaw in most resolvers dealing with hosts that don't exist: if a server says "I don't know shserver2" (rcode=3) the resolver will ask another one of your DNS servers.
It'll reduce DNS load if you put something in your zone for shserver2, I'd put 127.0.0.1.
-- Ned
____________
|
|
|
|
|
|
. . . Nice work there Matt / Berkeley yo 4 bruno ;)
Somebody break out a calculator and determine how many SETI@home classic clients this represents
. . . callin' Esme' ;)
|
|
|
|
|
|
So that is what was using the little bit of bandwisth, Classic just won't die. Interesting though. Facinating.
____________
BSG Anthem
My Facebook page
 |
|
|
mikeyVolunteer tester
Send message
Joined: 17 Dec 99
Posts: 4215
Credit: 3,474,603
RAC: 0
|
Here's a fun one: Since the switchover to using Hurricane Electric as our main ISP I noticed lingering traffic on the campus router which served our Cogent link. We're talking as much as 1 Mbit/sec. Today while updating lab-wide DNS records I noticed shserver2 was still there. This was the DNS alias for our SETI@home classic data server. I removed this entry, and check out the dip in traffic:
- Matt
Don't have a clue if it is relevant but waaaay back when, when you were having server issues ALOT of us put in actual server addresses in the "http proxy" section of Boinc. It worked until you caught on, seems we were using a connection line that cost you a fortune instead of a cheap one. I wonder if some hosts are still using that "technique".
____________
 |
|
|
|
|
So that is what was using the little bit of bandwisth, Classic just won't die. Interesting though. Facinating.
There’s still the occasional posting in the Q&A boards from people asking why their S@h v3.x doesn’t seem to be making any progress …
____________
  |
|
|
|
|
It worked until you caught on, seems we were using a connection line that cost you a fortune instead of a cheap one.
Another reason for the lot of you to make a donation. Payback.
____________
  |
|
|
mikeyVolunteer tester
Send message
Joined: 17 Dec 99
Posts: 4215
Credit: 3,474,603
RAC: 0
|
It worked until you caught on, seems we were using a connection line that cost you a fortune instead of a cheap one.
Another reason for the lot of you to make a donation. Payback.
Did so a month ago, sent a $100.00 donation thru the new paypal way.
____________
|
|
|
|
|
[snip]
On the plus side, we got all the pieces in place to move another function off kryten and onto bruno: file deletion. I just fired this off, and at first glance it seems faster. Time will tell if this is an improvement. Bruno is a faster machine in general, but kryten had a gigabit connection to the workunit file server, while due to lab infrastructure bruno can currently only have 100 Mbit. So we shall see. Hopefully queues will drain after we recover from the outage backlog.
[snip]
- Matt
Matt, now that Bruno is actually showing up on the "server" page, the "hosts" section of that page should be updated to include it, don't ya think?
____________
.
  |
|
|
Matt LebofskyVolunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: 1 Mar 99
Posts: 1375
Credit: 74,079
RAC: 0
|
Matt, now that Bruno is actually showing up on the "server" page, the "hosts" section of that page should be updated to include it, don't ya think?
Yes I do, which is why I did included it before I even started this thread. Fourth one down in the list. The list is in no specific order.
- Matt
____________
-- BOINC/SETI@home network/web/science/development person
-- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude |
|
|
|
|
Somebody break out a calculator and determine how many SETI@home classic clients this represents
Here's my response from the Café - can anyone improve the estimated input values or suggest a better model?
Well I'll start the ball rolling. There are two issues here - first a suitable model of the traffic origins and second, sensible input numbers.
From inspection of the graph I assume 0.7Mbit/S average shserver2 traffic, with 33% generated hourly and the rest generated daily. Reasonable choices of such values could affect the answer by factors of 2 or so and are thus not too critical for a first estimate.
From this conversation I will take an average packet size of 40 bytes (given that no actual data is being sent anywhere) This gives a little over 2000 packets per second and just under 200 million per day. I also assume that each contact results in 4 packets. But I really don't know in either case. The equality of input and output traffic rates suggests a strong minimum packet size effect.
Using a formula
numberofhosts = packetsperday/(hourlyfraction*24+(1-hourlyfraction))/packetspercontact
we then get 5768000 hosts (rounded a bit,) which at just under 6 million seems rather high, but it is more than a thousand and less than a billion, so it gives us something to talk about. |
|
|
Es99Volunteer tester
Send message
Joined: 23 Aug 05
Posts: 7339
Credit: 210,084
RAC: 269
|
. . . Nice work there Matt / Berkeley yo 4 bruno ;)
Somebody break out a calculator and determine how many SETI@home classic clients this represents
. . . callin' Esme' ;)
What am I? The secretary? :p
____________
 In the beginning the Universe was created. This has made a lot of people very angry and has been widely regarded as a bad move. - Douglas Adams |
|
|
|
|
. . . Nice work there Matt / Berkeley yo 4 bruno ;)
Somebody break out a calculator and determine how many SETI@home classic clients this represents
. . . callin' Esme' ;)
What am I? The secretary? :p
Secretary? NOPE - 'NOT' what i was thinkin' - i refer to Matt's Request which i have hope / belief - in the Positive Sense of the Word - that you may be able to assist with some 'Math' related (see Matt's Post) . . . Michael Roberts has started the 'ball rollin' per se . . . see IF it's something YOU could help out on . . . that's all (after all - YOU are GOOD with Math aren't you?) mi thinks so ;)
With Sincerity,
richard
|
|
|
|
|
. . . Nice work there Matt / Berkeley yo 4 bruno ;)
Somebody break out a calculator and determine how many SETI@home classic clients this represents
. . . callin' Esme' ;)
What am I? The secretary? :p
I was thinking "Executive Assistant".... <ducking>
____________
|
|
|
|
|
|
What's the significance of "Esme"? I see the green graph. Maybe somebody in this thread likes a Hispanic girl named "Esmeralda", which is Spanish for "emerald". Back in 1998 I saw parts of a Mexican soap opera "telenovela" on the TV channel "Univisión" with that title, starring Fernando Colunga y Leticia Calderón.
____________
|
|
|
|
|
What's the significance of "Esme"? I see the green graph. Maybe somebody in this thread likes a Hispanic girl named "Esmeralda", which is Spanish for "emerald". Back in 1998 I saw parts of a Mexican soap opera "telenovela" on the TV channel "Univisión" with that title, starring Fernando Colunga y Leticia Calderón.
Clyde - it's Es99 - 'er name is Esme' (i believe i 'ave the spellin' correct?)
|
|
|
|
|
Somebody break out a calculator and determine how many SETI@home classic clients this represents
Here's my response from the Café - can anyone improve the estimated input values or suggest a better model?
Well I'll start the ball rolling. There are two issues here - first a suitable model of the traffic origins and second, sensible input numbers.
From inspection of the graph I assume 0.7Mbit/S average shserver2 traffic, with 33% generated hourly and the rest generated daily. Reasonable choices of such values could affect the answer by factors of 2 or so and are thus not too critical for a first estimate.
From this conversation I will take an average packet size of 40 bytes (given that no actual data is being sent anywhere) This gives a little over 2000 packets per second and just under 200 million per day. I also assume that each contact results in 4 packets. But I really don't know in either case. The equality of input and output traffic rates suggests a strong minimum packet size effect.
Using a formula
numberofhosts = packetsperday/(hourlyfraction*24+(1-hourlyfraction))/packetspercontact
we then get 5768000 hosts (rounded a bit,) which at just under 6 million seems rather high, but it is more than a thousand and less than a billion, so it gives us something to talk about.
OK, call it 5 million connections. That does not equate to 5 million hosts. Assume that each host contacts the server 10 times per hour and you get around 25000 hosts instead. The 10 contacts per hour was a left handed extraction.
____________
BOINC WIKI |
|
|
|
|
Somebody break out a calculator and determine how many SETI@home classic clients this represents
Here's my response from the Café - can anyone improve the estimated input values or suggest a better model?
Well I'll start the ball rolling. There are two issues here - first a suitable model of the traffic origins and second, sensible input numbers.
From inspection of the graph I assume 0.7Mbit/S average shserver2 traffic, with 33% generated hourly and the rest generated daily. Reasonable choices of such values could affect the answer by factors of 2 or so and are thus not too critical for a first estimate.
From this conversation I will take an average packet size of 40 bytes (given that no actual data is being sent anywhere) This gives a little over 2000 packets per second and just under 200 million per day. I also assume that each contact results in 4 packets. But I really don't know in either case. The equality of input and output traffic rates suggests a strong minimum packet size effect.
Using a formula
numberofhosts = packetsperday/(hourlyfraction*24+(1-hourlyfraction))/packetspercontact
we then get 5768000 hosts (rounded a bit,) which at just under 6 million seems rather high, but it is more than a thousand and less than a billion, so it gives us something to talk about.
OK, call it 5 million connections. That does not equate to 5 million hosts. Assume that each host contacts the server 10 times per hour and you get around 25000 hosts instead. The 10 contacts per hour was a left handed extraction.
If there is a host at the IP, listening on the right port, you'd have six packets minimum per connection.
If nothing is listening, you'd have several "syn" packets per host that would go unanswered. I'd have to read a couple of RFCs to figure out the likely number.
____________
|
|
|
|
|
|
Well, here's some anecdotal evidence:
I probably have somewhere between 50 and 100 hosts still trying to get some classic workunits.
Those were from previous jobs, where I had SAH running as a windows service in the standard workstation build. I had permission to install those clients, but I was the helpdesk and I didn't want to receive complaints from users about their processor utilization, so I hid the process. It wouldn't suprise me if whoever is running IT at those places doesn't know or doesn't care about it. Those workstations will likely keep attempting to pull work until they're finally retired or rebuilt. |
|
|
