Possible win33 issue

hi folks

i already have an open ticket on this, though techs dont say there is an issue, so i would like input from the wider community on this -

Something weird is going on.

I have an order form on a site on win33, with an Access database in the backend.

It seems as though duplicate transaction are going through – but with different info. ie, email notification that was sent, says that row ID = NAME 1. but when i access the database now, for that same row ID , i get NAME 2.

The weird thing is that the autonumber – which should be unique – is evidently not! but thats impossible!!

so this is my theory -

Was this server by any chance in transit (to Waco) over the last few days?

If so,

I thinking that maybe some peoples DNS was sending them to the old server, but some to the new – therefore I am seeing the same ID numbers with different details each time.

Looking forward to valuable input!

thanks!!

there is a chance of some hours of cached dns yes, the same would be possible with account mover.
If you’d like we could zip of the MDB on each side and you could reconcile them maybe?

yes, that would be great

in fact i wouldn’t mind getting a zip of the MDB from each day since 20 April.

Ticket [RS #FZT-19074-487]

Dont know if each day for that period is possible, but I’ll have them take a look

Thanks, its quite critical,
because if there were DNS issues, this is the only way i have a chance in getting the information.

Hello,

We have just replied you in the Ticket [RS #FZT-19074-487].

great

it seems that was the issue indeed - there were at least 4 transactions that “got lost”

what can be done to prevent this issue in future transfers? maybe the transfer should be done strictly night time on weekends?

Doesn’t do anything to fix matter, as there is always someone having some ‘busy’ time. file migrations also must happen over a 4-5 day period as there is simply no other way for it to move such volume of data quickly.
It is taking an initial sync, then the updates of that data run more quickly by updating only the changes from there.

it really would not matter at which point it happens, there will be such possible on a flat file database.

That is why things like SQL are the best solution, yes SQL requires a definitive downtime for the move, they’ve ranged from 30 minutes to 2.5 hours at this point, but transactions aren’t lost in limbo land either, and the web server can see it on either side in the right form.

i hear you

Can someone show me the email notice that went out that announced the migration of Win33 ???

I have scoured my email, and I find none. I had a site that actually went down, because I didn’t realize it was on the schedule.

I bet if snooper got one, he would have been on top of his problem quicker as well.

jph,

We are not announcing every single server with a mass mail, we have announced that servers will be possibly every night and given forums and status link for them. (April 20th this was sent)

We cannot and will not be mailing every server on mass mailing updates. We can do server search mass mailings, but not with resellers clients as it returns results for ALL ours, yours, etc and resellers would not like us mailing their clients directly, so that option is a non starter.

Dude, you were doing so, and then you stopped I guess. waaaaaaaay back on April 1 or so, the announcement said:

In continuation to the datacenter move announced some time back, we are going to start the migration process for this Hsphere hosting cluster. This process will be done in parts, and you will receive emails like this, during the process.

…and then it went on to list the servers being done in that batch (not including win33).

Then when did you tell us “nevermind, you’re on your own to check the schedule every day”?

didnt get one either.

the migration list that is setup also seems to lag behind, until the transfer is already complete.

Yes, and the April 20 email said in the second sentence:

We have servers known as WinCF and Win22 ready to migrate services during a maintenance window tonight estimated from 10PM EDT until 4am EDT, or when completed fully. During this time control panel services will be unavailable.

And this gave the CONTINUED impression that each server was being announced in email.

Buried way down at the bottom, yes, it did provide the links for the status page, as had the first announcement. But nowhere did it say or imply that we would no longer be notified by email, at least sort of on a weekly/batch/warning basis.

We will email when there are major changes needed like for nameservers.
Otherwise it included this:

“We make every effort during the migration to ensure your sites are available and working, but there will be some errors the software makes, and some changes required for some sites to make them function.
We will also continue to migrate servers on a nightly basis as file sync and IP addressing permits the updates for such will be made on http://status.jodohost.com and our forums on http://support.jodohost.com will contain additional details of the progress.
In the coming days some Microsoft SQL servers will be migrated and require a downtime of 2-5 hours to complete, we will strive to provide at least 12 hours advanced notice when this will happen on the above listed URLs, these are important to move with the servers most using them as much as it is possible so that you do not have to utilize remote SQL for an extended period of time.”

We are NOT going to stop mailing we are just not mailing for every migration.

You are have received updates emails, and there will be more, but you will not be getting 30+ emails for server migrations. Sending an email commits us to proceeding even if the situation is not right, where the forums and status page allows us to more easily call it off if required which has happened twice.
The Hsphere tools are not perfect, and they make some big msitakes, we can’t proceed if we detect them in finalizing the process.

you will likely get 2-5 more still through the process.

Stephen,

I completely understand your issues with the announcement emails, but you have to admit this was not clear from the announcements so far. Look at it from my side (and the I am sure I am not alone).

I am a small reseller, i only have a few servers to keep an eye out for. I see announcements coming in that list the servers that are scheduled for migration, great, I look to see if mine are listed. There is nothing in those emails that says the listing of specific servers will be inconsistently included in the announcements.

Now, ok, I hear that is the policy: inconsistency.

So, now my question is: Is there a single announcement thread that is dealing with migration announcements so that I can subscribe to THAT, and will that have advanced warnings, or will it only give status as things are happening or maybe delayed?

Twitter?

It is linked on every post of mine.

Here is an RSS feed for non twitter users quick reference, can add it to you ie/firefox/chrome quick feeds:

http://twitter.com/statuses/user_timeline/132992283.rss

I am really not going around looking for reasons to *****, but if you are NOT going to maintain a tool, and it is going to show false information, you should just take it down. Right now I understand win17 and win18 are in migration progress.

but…
http://status.jodohost.com/migr/index.php/MigrMain/index
says:

win17.mhere.biz Not Available Not Started
win18.m
here.biz Not Available Not Started
win19.m****here.biz 2011-06-12 DATA SYNC IN PROGRESS

and when I try to “look up my new IP”, i put in old IP’s for win17:
64.187.109.241
64.187.109.242
and get:
No Data requested or found

I don’t know. I don’t think I am really, really, really stupid, but maybe.