Ned

Not trying to be argumentative. Anyway you look at it, it will be tough...

The problem, notifying 5.4 million users that Seti Classic is dead...
IF UCB sends 5.4 million emails, they are a spammer
IF UCB sends 5.4 million emails, they may get someone that stopped doing Seti to start BOINC again
IF UCB sends 250,000 emails to registered "active" email addresses
* some will get through the user monitors and is still active
* some will get through the user is already migrated or running both Seti Classic and Seti BOINC
* some will get through the user is in auto and now has information
* some will bounce, the user already has a New Seti BOINC account
* some will bounce, the user has a new email address and "little means" to correct the old email
* some will bounce, the user quit ages ago, does not care

Actually, by definition, the E-Mail from SETI would not be unsolicited or commercial, so it fails all "sane" definitions of SPAM.

Soliticted Yes, I did not "sane," the gateway I built a year ago would see all the connections and then set it aside and refuse additonal connections. So irrespective of whether the user name is valid or will be bounced. There are connection thresholds... The gateway is just following rules, and everybody has their own definition of "sane"...
* valid DNS for the sending domain
* valid return address
* connection requests/sec
* number of names in the CC/BCC field
for a few


But SETI should concentrate on the 250,000 who are active with Classic as the rest of the 5.4 million are essentially gone now.

Once the 250,000 are notified, I'd start (slowly) inviting those who disappeared to come back. If most messages bounce, and few return, then it isn't worth the time and machine cycles, and I'd stop.

I'd also set up computers and domains just for this mailing, so that they could be blacklisted without affecting general campus mail.

As pointed out many times, the project can only do things that are possible.

Actually for the 250 thousand or less doing a seperate server is not a bad idea but I would expect it to come from Seti. IF the target mail server has DNS checking/blackhole turned on a Bogus Domain will be stopped/reported... then tracking which user got the email and which was blackholed would be tougher.

IF the mail is sent is small mixed groups (domains), and/or outbound connections are throttled (10 connections/sec or less); then the target mail domains should not have many issues. So what it takes a full 24 hours or couple of days it easier on the infrastructure...

So as much as the "sender" would like to open a connection to somedomain.com and deliver 10,000 messages... In this day and age it probably would not fly...

As I think about it you could use an SMPT engine, with the Message the list of names and a script to throttle the send operation... As long as it has a valid From address and the host has a proper DNS record, then the person sending could actually "monitor/log" the progress without having to have the postmaster account. This could be done from a workstation and paused if desired

I have used Blat to do monthly tests for specific keywords etc... and can limit the number of connections.

---

Please consider a Donation to the Seti Project.

Lee

Very good point, as to overiding preferences...

I do have a friend that has two Classic email accounts and would like them merged as it would affect his final "status in the world." Yes the forgotten password.

* They have not received notification that something is about to happen, which would allow them to decide

Realistically, when it is about to happen, UCB can take the email database of the users that are still returning WU's within the last year and compare it to the "migrated" accounts... IF the email addresses match, do not send an email... That will reduce the 250,000 by xx.xx percent. In that UCB is being responsible in email ethics...

UCB could create a small database and offer for users to enter Old Email Address and New Email Address and remove those for the quarter million to send, of course some validation would be required. Then transfer information to the new BOINC account... Yes, it would the requires some labor.

also the option in each users account for "send me email?" should be honoured as well, after all, that's what it's for ;) obviously something as major as this might qualify overriding that preference

* They can not carry "user" information because of the "bad email account"

UCB has already stated that, if you do not remember your password and we have no way for the computer to send it to you "create a new account."

another suggestion for the "old email" problem, why not do as CPDN have done, and under project prefs there is a field where they can enter their classic username to have their stats linked to their new boinc account, obviously UCB will probably come up with improvements on this in some way, maybe being able to enter your "old" email address to make things easier?

So yes, there are things to consider

---

Please consider a Donation to the Seti Project.

Ned

Not trying to be argumentative. Anyway you look at it, it will be tough...

The problem, notifying 5.4 million users that Seti Classic is dead...
IF UCB sends 5.4 million emails, they are a spammer
IF UCB sends 5.4 million emails, they may get someone that stopped doing Seti to start BOINC again
IF UCB sends 250,000 emails to registered "active" email addresses
* some will get through the user monitors and is still active
* some will get through the user is already migrated or running both Seti Classic and Seti BOINC
* some will get through the user is in auto and now has information
* some will bounce, the user already has a New Seti BOINC account
* some will bounce, the user has a new email address and "little means" to correct the old email
* some will bounce, the user quit ages ago, does not care

Actually, by definition, the E-Mail from SETI would not be unsolicited or commercial, so it fails all "sane" definitions of SPAM.

Soliticted Yes, I did not "sane," the gateway I built a year ago would see all the connections and then set it aside and refuse additonal connections. So irrespective of whether the user name is valid or will be bounced. There are connection thresholds... The gateway is just following rules, and everybody has their own definition of "sane"...
* valid DNS for the sending domain
* valid return address
* connection requests/sec
* number of names in the CC/BCC field
for a few


But SETI should concentrate on the 250,000 who are active with Classic as the rest of the 5.4 million are essentially gone now.

Once the 250,000 are notified, I'd start (slowly) inviting those who disappeared to come back. If most messages bounce, and few return, then it isn't worth the time and machine cycles, and I'd stop.

I'd also set up computers and domains just for this mailing, so that they could be blacklisted without affecting general campus mail.

As pointed out many times, the project can only do things that are possible.

Actually for the 250 thousand or less doing a seperate server is not a bad idea but I would expect it to come from Seti. IF the target mail server has DNS checking/blackhole turned on a Bogus Domain will be stopped/reported... then tracking which user got the email and which was blackholed would be tougher.

IF the mail is sent is small mixed groups (domains), and/or outbound connections are throttled (10 connections/sec or less); then the target mail domains should not have many issues. So what it takes a full 24 hours or couple of days it easier on the infrastructure...

So as much as the "sender" would like to open a connection to somedomain.com and deliver 10,000 messages... In this day and age it probably would not fly...

As I think about it you could use an SMPT engine, with the Message the list of names and a script to throttle the send operation... As long as it has a valid From address and the host has a proper DNS record, then the person sending could actually "monitor/log" the progress without having to have the postmaster account. This could be done from a workstation and paused if desired

I have used Blat to do monthly tests for specific keywords etc... and can limit the number of connections.

Pappa,

I run a small ISP. I've got this stuff down cold.

The domain needs to be seti-related. It could be something like setiathome.org if that's available, or it could be news@setiathome-news.ssl.berkeley.edu.

I know of no mail server that does not log. It's a one time mailing, there isn't much to learn from the bounces. I wouldn't be surprised to find that half of the 250,000 addresses are just plain bad. You might as well just discard anything that comes in.

The return address should be valid. It doesn't have to go to a mailbox. The postmaster address (and abuse addresses) should only get complaints, not real mail. Maybe 25,000 "thanks but no thanks" messages and the rest just trash.

It'd be nice to contact those folks, but the work involved in sorting is unreasonable.

If I worked at Berkeley, it'd be an old Sun server or maybe BSD Unix running SendMail.

---

To clarify , does this mean that if I receive a unit on the 14th , and send back the result on the 15th it will be the last ; or will I receive my final classic unit on the 15th ?

---

Hi

The last result will be send out before the shutdown.
Maybe a week or so.

greetz Mike

---

With each crime and every kindness we birth our future.

Greybeard the message I was responding to had NOTHING to do with the closing of Classic so pls don't single me out!

Well, you're right in a certain sense... the hijack could have begun in that post already...

[...]

:-)= Greybeard

Hey, that was my post he was responding to - and I wasn't starting a "Classic vs BOINC"-thing, I was merely pointing out that the 2-step getting started info on the main page is misleading. A fact that seems to be confirmed by the post I responded to, suggesting a guide for migrating to BOINC.
Why ask for a migration guide if the 2-step guide on the main page is OK?

---

"Are you suggesting coconuts migrate?"

Greybeard the message I was responding to had NOTHING to do with the closing of Classic so pls don't single me out!

Well, you're right in a certain sense... the hijack could have begun in that post already...

[...]

:-)= Greybeard

Hey, that was my post he was responding to - and I wasn't starting a "Classic vs BOINC"-thing, I was merely pointing out that the 2-step getting started info on the main page is misleading. A fact that seems to be confirmed by the post I responded to, suggesting a guide for migrating to BOINC.
Why ask for a migration guide if the 2-step guide on the main page is OK?

PEACE! I surrender!
:-)= Greybeard

---

All about BOINC: BOINC-Wiki (by Paul D. Buck)