Scheduled Maintanence cp.hsphere.cc

[Silverbug]

Yeah if it was run on a saturday, at least then it would be the weekend everywhere else(well i assue it would)

lol i dont really know why this bothers me, as my clients dont even have access to the cp anyway lol

[tmannino]

I read through the post and I think this much down time is unacceptable. I several webservers and database servers setup as cluster boxes so my machines are never down no matter what maitenance I need to do. Why are these machines not clustered with redundant Database servers?

Tim

[admin]

The reaosn H-SPhere is not a clustered server yet is because psoft does not support it yet. We are working on doing a clustered pgsql box our self but when you run a vaccum on pgsql no mater how big the cluster is you still show stop anything from accessing it or it will not fully finish. You can read this under the vacuum section right on pgsql's site because this would still be changing in the DB.

We are making things redundant where we can with in our price point, but I can already tell you we are past most other H-Sphere hosts in how adv. our setup already is. I can only think of even one other host that has a server to just run pgsql for the CP, I can't think of any that has as many servers or customers as we do. If we make every thing redundant down to the servers this would drive our cost up x 2 and would drive our prices up by 2. So $35 would jump to $70. Thats not really the market this cluster was built and priced out at.

Think about it this way as well. We have 35 windows servers and 16 unix servers plus all the others, we might have 1 big issue every 30 days for all the servers being it hardware, some months worse or better. If you have grown with us since we started then most of your clients are spread over lots of systems and when there is an issue you get a LOT less calls. If you were on a CPanel box when it goes down all sites go down. H-Sphere is 10x better than the single system installs already with out the redundant hardware.

Thats just my 2 cents on it. I would love to be able to make all the systems clustered with redundant servers including web servers but with the lack of funds it becomes hard. You have to also take in to acount we are not using low end hardware ether, we use dual xeon, 1/2GB of ram full scsi systems. Not the cheap P4 systems with SATA like lots of others do.

[generic]

Originally Posted by tmannino

I read through the post and I think this much down time is unacceptable. I several webservers and database servers setup as cluster boxes so my machines are never down no matter what maitenance I need to do. Why are these machines not clustered with redundant Database servers?

Tim

If you have several webservers and database servers, why are you here?

[antic]

Well, I'm totally happy with Vortech's setup, except perhaps on one point, which is they don't have an Oz branch.

Anyway, how many other hosts communicate this much and openly with customers and even consider the possibility of changing things, like running db vacuums on the weekend?

(did you detect the subtle supportive encouragement there?)

[jetzkr8]

Originally Posted by admin

We have 35 windows servers and 16 unix servers plus all the others, we might have 1 big issue every 30 days for all the servers being it hardware, some months worse or better. If you have grown with us since we started then most of your clients are spread over lots of systems and when there is an issue you get a LOT less calls. If you were on a CPanel box when it goes down all sites go down. H-Sphere is 10x better than the single system installs already with out the redundant hardware.

Agree on the technology part. Not many hosting companies can offer the same thing. No HSphere provider is close, from what I have noted.

In any case, I have one suggestion. When the CP is down for maintenance, there is an erro message or no message. It would be nice to have some form of standard screen indicating the disruption is a scheduled maintenance job and temporary in nature. It would further help if some ETA is put up as well. This way most people/customers trying to come in are informed (and understand) while also not trying consistently to see if things are already up. Save everyone a lot of time and make it convenient.

[jetzkr8]

Originally Posted by admin

.......we use dual xeon, 1/2GB of ram full scsi systems. Not the cheap P4 systems with SATA like lots of others do.

By my last count, there were soem PIII boxes, but the page showing server configs has been taken off the main website. Can you confirm all servers are as stated above?

[admin]

Originally Posted by jetzkr8

Agree on the technology part. Not many hosting companies can offer the same thing. No HSphere provider is close, from what I have noted.

In any case, I have one suggestion. When the CP is down for maintenance, there is an erro message or no message. It would be nice to have some form of standard screen indicating the disruption is a scheduled maintenance job and temporary in nature. It would further help if some ETA is put up as well. This way most people/customers trying to come in are informed (and understand) while also not trying consistently to see if things are already up. Save everyone a lot of time and make it convenient.

We use to have it say something like it was down for maintenance any time the system was down for anything. I will email James and Dean and see if they can't fix that so it does it again.

I hope this does not take 1/2 a long this time as we have really cleaned up the DB a LOT since last month.

[Dean]

The CP is down for the vacuum right now. I have modified the index.html to reflect the 'nice' error message.

[Dean]

Just to address the current flood of tickets, if your browser has cached the old index, it will still try to direct you to the java servlet...which will of course show the error. Only way to fix that on our end is to edit PSoft's code. For anyone that has actually seen that codebase, I think you will agree that it is not worth it.

[mresell]

I do think that on the weekend would possibly be better, but really how vital is it that the cp be up 24/7.? A few hrs. out of a month is not bad. Are people worried about signups? Cause no one should need to be in the cp that much. (altho redundancy would be nice, but hsphere is what it is.) I would think that down for maintenance looks better than a slow or unresponsive cp tho.

[generic]

Originally Posted by Dean

The CP is down for the vacuum right now. I have modified the index.html to reflect the 'nice' error message.

Internal Server Error
The server encountered an internal error or misconfiguration and was unable to complete your request.
Please contact the server administrator, root@cp.tracewebspace.com and inform them of the time the error occurred, and anything you might have done that may have caused the error.

More information about this error may be available in the server error log.

Additionally, a 404 Not Found error was encountered while trying to use an ErrorDocument to handle the request.

I guess that is nice

Can we get an ETA on when this will be finished?

[Dean]

The vacuum is complete and the CP is back up.

[cambodia]

mean the schedule mainternance is 100% completed ?