Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!


GoodHosting.co | Upcoming Maintenance Windows
New on LowEndTalk? Please Register and read our Community Rules.

All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.

GoodHosting.co | Upcoming Maintenance Windows

Hello LowEndTalk,

Pursuant to our previous maintenance window:

August 25th, 2014 ; 06:00 PM PST - 07:00 PST (1 Hour): We'll be taking down the networking for our main website and a few other internal services for approximately an hour to replace one of the fabric units in our Chicago cluster. Unfortunately, we cannot use the failover unit due to the port configuration. Something we will be fixing in the same window. This is a preventative maintenance to prevent this from being an issue in the near future, as the network status shown by the BladeManager is currently degraded due to this fabric unit.

We will be conducting the following follow-up maintenance windows:

August 26th, 2014 ; 05:00 PST - 07:00 PST (2 Hours): We will be taking down networking for our main website and our entire internal cluster framework, including the OpenNebula access panel for the Chicago cluster for approximately two hours to replace one of the fabric units in our Chicago location. We were unable to successfully replace the fabric unit during the previous day's maintenance window, as the firmware and configuration offload from the neighboring VCM unit took the entire time.

>

During this maintenance window, the Chicago1 OpenNebula panel will be inaccessible, as well as our primary DNS servers and our website / billing panel. The other OpenNebula locations will be accessible only via IP address, so please remember to use this instead of the DNS name during this time.

>

The BMs are currently reporting a network "Degraded" status for our primary 10Ge fabric interconnects due to this errant module. The very same module from yesterday is currently still in production, and will be removed and swapped out during this maintenance window. The incremental configuration transfer, and VLAN/MAC/ROUTE table update propagation from the failover switch should only take twenty minutes.

>

Diminished network capacity and higher latencies will be experienced for a period of at least four hours after the maintenance window concludes, as the failover unit and the primary fabric fiber unit must synchronize their configuration, VLAN table, MAC cache, ROUTEing table, and various other settings. The VCM will also take some time to module update the software on the fabric switch. During this time, the failover switch will be acting as the primary switch; but at a diminished processing capacity to speed up the transfer.

Comments

  • So what? I'm not your client. Send an email to them, not us...

    Thanked by 1raindog308
  • @cassa people bitched previously that he didn't notify customers publicly about the website outage, i guess that's the reason for this post.
    It's impossible to win in such situations - some people will complain that he didn't post, some people will complain that he did post.

    Thanked by 1hostnoob
  • Good job :)

    Thanked by 1linuxthefish
  • wychwych Member

    It's impossible to win in such situations - some people will complain that he didn't post, some people will complain that he did post.

    Damned if you do, damned if you don't ;)

  • Not your client, but good job alerting your clients on LET.

    I just hope that not every company starts doing this. An occasional post about this is kinda nice, tho.

    Thanked by 1linuxthefish
  • rds100 said: It's impossible to win in such situations - some people will complain that he didn't post, some people will complain that he did post.

    I don't think anyone who is not a client will complain when they use their usual methods of notification (i.e. emailing all current clients + announcement on Twitter or whatever)

  • This kind of proves what I was saying that LET won't ever be happy with how you handle something, whether you follow reporting procedures or not (or if you're following LET advice or not) you're still damned. Damned about having too many DNS servers when I didn't have enough, damned about posting maintenance windows in the section specifically designed for provider announcements such as maintenance windows and downtimes, damned when I don't...


    @Gadelhas said:
    I just hope that not every company starts doing this. An occasional post about this is kinda nice, tho.

    This is apparently what the provider section is meant for, maintenance windows that are large enough to take a significant portion of service (including our website, that being the defining factor) offline. If this only affected a few servers, there would be no need to post it here; as it wouldn't affect enough people.

    Thanked by 1Pwner
  • wychwych Member

    @GoodHosting, chill man! Its only Tuesday...

    But if it makes you feel better most people here can relate in some way.

  • @wych said:
    GoodHosting, chill man! Its only Tuesday...

    :'(, but I want it to be Friday again already, dealing with HP networking is a serious pain. These fabric modules aren't even real switches, they don't have any of the features you expect from a switch, and literally all traffic has to be VLAN tagged at the OS level, the fabric ports do not support VLAN tagging on entry, and they do not untag (but they refuse to let non-tagged packets leave the enclosure, and reach our actual switches.)

    It's a nightmare. + The web interface only works in Internet Explorer 6 on Windows XP, it doesn't load in Chrome or Firefox, nor in any newer version of Internet Explorer. The CLI also doesn't even work in their newest firmware revisions, and the firmware is a pay-to-download software from their website (but on a good note, the hardware is supported until 2025.)

    But I digress, all of this is unimportant ranting on my part.

  • GoodHosting said: The web interface only works in Internet Explorer 6 on Windows XP

    Someone needs to make a reverse modernizr/ie9.js/ie7.js - something that makes old, crappy JS powered websites work in modern browsers (rather than make modern websites work in old browsers).

    That's shit though, you should dump it if you can. I couldn't even use that UI if I wanted, I don't have Paralells installed or anything any more and I don't really want to either.

  • mikhomikho Member, Host Rep

    @GoodHosting said:

    You must be using really old hardware or the wrong model for what you expect it to do.

  • Will you be announcing your next holiday here too?

  • GoodHosting said: The web interface only works in Internet Explorer 6 on Windows XP

    Java? The CLI should work, and most procurve (if it is) stuff has lifetime warranty now and excellent support!

  • @rds100 said:
    cassa people bitched previously that he didn't notify customers publicly about the website outage, i guess that's the reason for this post.
    It's impossible to win in such situations - some people will complain that he didn't post, some people will complain that he did post.

    Then again, I think an e-mail would suffice. I personally think that if people receive an e-mail and don't read it, they have no right to complain as they could have known.

    I'm letting this be, though :-)

    Thanked by 1Dylan
  • @mpkossen

    OVH being listed a source of attacks and spam = recieving e-mail? I guess you know how it is, right? The chances are some clients still won't get the email as their mail provider has blocked OVH.

    For that mandrill or something similar though ;).

  • Thank you for update here. client email changed for better communication (:

Sign In or Register to comment.