I just fixed the core monitoring to follow the HAProxy instance when the load balancers fail over but the node name is always "fw1" because the configs are identical. Stupid monitoring system exposing the flaws in my grand schemes!
Good Morning Ed – I had a little discussion with Win about the interesting stuff around Linux RAM caching. Would it help PPro performance if we added more RAM to the PPro vm for caching, or are we barking up the wrong tree here? We have the resources sitting here idle, and if it would help, we can add more RAM. Thoughts?
@ewwhite Okay, good. I know what it's like to go into deep detail about how to do things right, but then have the recipient wipe their butt with it because 1) Too expensive, or 2) Because I want what I want... and on top of it, you're not charging for time for one reason or another. I can handle disrespect and stupidity if I'm getting paid my hourly rate. =P
No, the RAM is sized optimally for your setup. The Linux VFS cache layer is for reads, not writes.
I’ve clearly outlined what’s necessary to achieve better performance. It is not possible with your current infrastructure.
dude these things are freakin' metal. The box I'm working on had two disks fail in a volume group, a filesystem (THE FILESYSTEM with all THE DATA) had 918 extents sitting with both copies on those disks. They lucked out. The filesystem fscked. I forced the VG and FS online. No data loss.
@ewwhite this is their legacy mostly read-only ERP system with 40-ish years of records. Anyone who know how to AIX is long gone. Turns out the "backup" they thought they were making was just a mksysb (bootable system restore of the root volume group (yes, these fuckers can BOOT FROM TAPE to restore)).
@voretaq7 oh right, but of course I knew in this case it was :P
@ewwhite Right - but maybe I'm being to opportunistic about it, because I have a clean conscience with continuing to charge for time after I very clearly stated "You're doing it wrong and asking me to advise you on how to do it wrong in the rightest way possible. You know I'm billing for this? You're okay with that? Okay, well..."
So that would keep me on the fishing hook, but you do things differently and have, you know, success. So it's different for you.
@ewwhite IMO it's best to be politely abrupt. So just "My recommendation for bare metal will solve these problems, and you will still be highly available and highly recoverable. VMware confers no advantages that can't be matched by my solution, and only adds downfalls. In the interest of narrowing my liability, I'll let my existing recommendations stand. If you'd like to pursue this VMware trouble to a satisfactory solution, I can recommend the following consultants:"
Cut the cord firmly and cut it as soon as you feel squeemish about the situation.
The longer it bleeds out, the weaker your position becomes.
Actually I'm so busy right now... imaging, Samba work, Samba work, Discourse, beating people over the head with RFCs, trying to explain to a company why they shouldn't be relying on 5-year old crummy hardware for core business functions, the usual.
@BigHomie Nothing special...it's a simple C# app I wrote. Build -> Build Solution ...I then manually run the script and it copies the files. The files copy while the drive is mounted now (robocopy was failing, but regular "copy" works fine) but it won't disconnect
Has anyone figured out how to attach a programmable electrode to a mouse and/or keyboard yet? I need to start delivering triggered, programmable electric shocks to the users.
This happened to me yesterday.
Someone claimed she sent me an e-mail, but I did not got it.
It's not in Inbox, Spam, Trash, etc.. it's nowhere.
How is this possible (assuming there is no user error)?
Where can things go wrong?
@MattBear Okay, I thought maybe SpiceWorks sent you an email when you got a PM. That's the only other place I knew how to direct contact you since SF/SE doesn't show email addresses and neither does Fitocracy.
I did this at a previous job, so here are a few tips for what will make it easier on everyone. Plan at least 4, preferably 6, weeks out. This gives your watch-standers the chance to arrange their lives around...
@MattBear That's nice. I hit 195-ish late last year but got nailed with vertigo, couldn't work out, then got lazy, so I just decided to start cutting down, leaning out and then rebuild the muscle. I'm about 173 now. Aiming for 165 before I start the bulk up again.
Sigh... you see man, it's all about how you present it:
Tough shit for them. Also, you're not putting a router in their office/home. You're putting a Phone Connectivity Device. The only thing behind it should be the phone. — MikeyB22 mins ago
@MikeyB does that let you get frontend/backend status information, connections, etc? Or do you just monitor the frontends via whatever protocol and alarm if they go down?
Use the unix socket directly: cat <<EOF | socat stdio /var/run/haproxy-1.5.stat | grep -v ^$ disable server app/app${BE}_00 disable server app/app${BE}_01 disable server app/app${BE}_02 disable server app/app${BE}_03 EOF
:15282250 um.... the core of the script is:
while(1) {
my ($client_socket, $haproxy_socket, $cmd, $res);
# waiting for new client connection.
$client_socket = $socket->accept();
while ($cmd = <$client_socket>) {
$haproxy_socket = new IO::Socket::UNIX (
Type => SOCK_STREAM,
Peer => '/var/run/haproxy.sock'
) or warn "No HAPROXY Socket!";
$haproxy_socket->send($cmd);
while ($res = <$haproxy_socket>) {
$client_socket->send($res);
}
$haproxy_socket->close();
I want to write a server that can accept multiple clients in python (twisted). I am already quite familiar with socket programming with the standard python socket module but here comes the trouble..
I think twisted is really hard to get into and i have read some tutorials about it.
But a thing t...
@voretaq7 No, but I run an email appliance that does inbound mail handling for a few hundred domains.
@voretaq7 All inbound SMTP connections are run through a gauntlet of three RBLs, which is why queries are so high. You gotta get through SpamCop, SpamHaus, and Barracuda to make it through.
@Wesley my access map has a bunch of hard-fails in it that say "I unsubscribed from your shit, you didn't honor the unsubscribe. You're never sending mail to bsd-box.net again."
(that's mainly for the ones that SA doesn't catch)
@voretaq7 Actually my RBL rejection messages are tame. I thought about being mean, but... sometimes it's just too much work to be mean. I simply drop a link to the proper dispute page for the RBL. "Here, go deal with it you poop flinger."
my "You're getting around the anti-spam measures, and you annoy me" messages are basically "It's my fucking server, and I'm not letting it talk to you anymore."
The settings are found within Windows Backup's utility itself.
http://windows.microsoft.com/en-us/windows/set-change-automatic-backup-settings#1TC=windows-7
The scheduled task does exist though. It's under Scheduled Tasks, Task Scheduler Library, Microsoft, Windows, WindowsBackup.
@HopelessN00bGeniusofnetwork - I voted for you in the election, but it appears you have around 130 dickfers.
This happened to me yesterday.
Someone claimed she sent me an e-mail, but I did not got it.
It's not in Inbox, Spam, Trash, etc.. it's nowhere.
How is this possible (assuming there is no user error)?
Where can things go wrong?
The settings are found within Windows Backup's utility itself.
http://windows.microsoft.com/en-us/windows/set-change-automatic-backup-settings#1TC=windows-7
The scheduled task does exist though. It's under Scheduled Tasks, Task Scheduler Library, Microsoft, Windows, WindowsBackup.
