O U T A G E R E P O R T F O R M ===================================
Incident Date: 6/28/07
Incident Time: 6:00PM EDT
Elapsed Time of Incident: Ongoing
People/Groups Impacted: Package builders / update tool / koji web services
Site Affected: koji.fedoraproject.org
Description: The koji.fedoraproject.org server is experiencing very high load. Attempts to login and address the load issue are under way.
Future Recommendations: Continue investigating process issues with httpd (if that proves to be the culprit once more).
On Thursday 28 June 2007 19:46:22 Jesse Keating wrote:
O U T A G E R E P O R T F O R M
Incident Date: 6/28/07
Incident Time: 6:00PM EDT
Elapsed Time of Incident: Ongoing
People/Groups Impacted: Package builders / update tool / koji web services
Site Affected: koji.fedoraproject.org
Description: The koji.fedoraproject.org server is experiencing very high load. Attempts to login and address the load issue are under way.
Future Recommendations: Continue investigating process issues with httpd (if that proves to be the culprit once more).
Outage is now over.
Future Recommendations: Properly document location of koji server and instructions on how to power cycle. Grant access to admins in multiple timezones for "round the clock" coverage.
On Friday 29 June 2007 09:56:33 Till Maas wrote:
Can you please change the date format to ISO-style: e.g. 2007-06-28 and the time to 24h, UTC time?
Sure, why not. I kifed this mail from our internal outage reports which report things in HQ time. We don't have a "formal" outage template yet in Infrastructure, I had wanted to do that at the last meeting but I was otherwise busy.
Jesse Keating wrote:
Sure, why not. I kifed this mail from our internal outage reports which report things in HQ time. We don't have a "formal" outage template yet in Infrastructure, I had wanted to do that at the last meeting but I was otherwise busy.
Actually we do :)
http://fedoraproject.org/wiki/Infrastructure/OutageTemplate
-Mike
On Friday 29 June 2007 10:53:49 Mike McGrath wrote:
Jesse Keating wrote:
Sure, why not. I kifed this mail from our internal outage reports which report things in HQ time. We don't have a "formal" outage template yet in Infrastructure, I had wanted to do that at the last meeting but I was otherwise busy.
Actually we do :)
http://fedoraproject.org/wiki/Infrastructure/OutageTemplate
-Mike
That's for planned outages. I think we need one for unplanned outages that is organized differently to get the pertinent information to the developers as quickly as possible.
On Friday 29 June 2007 11:34:10 Jesse Keating wrote:
http://fedoraproject.org/wiki/Infrastructure/OutageTemplate
-Mike
That's for planned outages. I think we need one for unplanned outages that is organized differently to get the pertinent information to the developers as quickly as possible.
Although looking at the current template that may work just fine (: I bet we could even create a tool that asks for input and spits out a properly formed email.
On Friday 29 June 2007 9:56 am, Till Maas wrote:
On Fr Juni 29 2007, Jesse Keating wrote:
O U T A G E R E P O R T F O R M
Incident Date: 6/28/07
Incident Time: 6:00PM EDT
Can you please change the date format to ISO-style: e.g. 2007-06-28 and the time to 24h, UTC time?
Regards, Till
+1 to use ISO-style, but please list UTC *and* local time for the outage.
On Fri, 29 Jun 2007 10:09:55 -0400, Jeff MacDonald scripst:
+1 to use ISO-style, but please list UTC *and* local time for the outage.
What local time? Why should EDT matter to me or other people living somewhere else than on the eastern shore of the United States (and maybe in Brazil, I am not sure)?
Matěj
On Friday 29 June 2007 10:30 am, Matej Cepl wrote:
What local time? Why should EDT matter to me or other people living somewhere else than on the eastern shore of the United States (and maybe in Brazil, I am not sure)?
Matěj
because not everybody knows what UTC/GMT is, and using it alone will just cause further confusion.. if, however, one can see that UTC is a few hours off of US/Eastern, things might go better. :)
at the moment it sounds like the outage notification is being generated by hand, so listing the two different time zones will take a little extra work.. however, if the process is automated, it will be trivial to make a call to "gmtime()" as well as "localtime()".
please.. if this issue causes the notifications to *not* be sent, leave it the way it is.. I'd rather get the notifications and have to do math to convert to my local time than not get them at all because there is no agreement on which time zone(s) should be used.
regards,
On Fr Juni 29 2007, Jeff MacDonald wrote:
because not everybody knows what UTC/GMT is, and using it alone will just cause further confusion.. if, however, one can see that UTC is a few hours off of US/Eastern, things might go better. :)
I do not know, what EDT is, either. But I know the offset of my local time to UTC.
at the moment it sounds like the outage notification is being generated by hand, so listing the two different time zones will take a little extra work.. however, if the process is automated, it will be trivial to make a
Maybe it is enough to mention the offset of EDT to UTC, so everyone in EDT can see what he has to calculate.
please.. if this issue causes the notifications to *not* be sent, leave it the way it is.. I'd rather get the notifications and have to do math to convert to my local time than not get them at all because there is no agreement on which time zone(s) should be used.
Even with UTC I still have to convert the time to CEST, but UTC is imho a fair common timezone.
Regards, Till
On Friday 29 June 2007 10:57 am, Till Maas wrote:
On Fr Juni 29 2007, Jeff MacDonald wrote:
because not everybody knows what UTC/GMT is, and using it alone will just cause further confusion.. if, however, one can see that UTC is a few hours off of US/Eastern, things might go better. :)
I do not know, what EDT is, either. But I know the offset of my local time to UTC.
EDT is "US/Eastern Daylight Time". I should have worded the original reply a bit differently.. not all *Americans* know what UTC/GMT is.. I'm odd because I worked for ans.net for a while, and we did a lot of things using UTC.
there's been some replies to this thread showing an easy way to convert to local time (assuming the machine is configured correctly), so I'm ok with the template that was created that uses UTC.
at the moment it sounds like the outage notification is being generated by hand, so listing the two different time zones will take a little extra work.. however, if the process is automated, it will be trivial to make a
Maybe it is enough to mention the offset of EDT to UTC, so everyone in EDT can see what he has to calculate.
that works for me.
Even with UTC I still have to convert the time to CEST, but UTC is imho a fair common timezone.
which is why I thought it best to list UTC *and* local time to the equipment in question so that it would be easy for the vast majority of users to figure out what happened and when.
Regards, Till
regards,
Jeff MacDonald wrote:
EDT is "US/Eastern Daylight Time". I should have worded the original reply a bit differently.. not all *Americans* know what UTC/GMT is.. I'm odd because I worked for ans.net for a while, and we did a lot of things using UTC.
Not all Americans are Fedora packagers. I'd say the types of people that package for a Linux distribution would be more likely to know what it is. Or would at least know to google for it. :-)
On Friday 29 June 2007 11:16 am, Christopher Aillon wrote:
Not all Americans are Fedora packagers. I'd say the types of people that package for a Linux distribution would be more likely to know what it is. Or would at least know to google for it. :-)
ok.. the issue here is to save time/effort. I get several hundred non-junk emails during the work week. if I have to do a google search or open a terminal to figure out if an outage affected me or not, I might just skip it or get the conversion wrong.
if the notifications are not automated, then by no means should an engineer/packager/whatnot be expected to list UTC *and* a local time zone.
let's keep it at UTC for now, and *please* use the ISO date format per the template.
hey.. at least notifications are being sent out. :)
regards,
Le Ven 29 juin 2007 17:10, Jeff MacDonald a écrit :
Even with UTC I still have to convert the time to CEST, but UTC is imho a fair common timezone.
which is why I thought it best to list UTC *and* local time to the equipment in question so that it would be easy for the vast majority of users to figure out what happened and when.
??? The vast majority of users is not in the same timezone as the equipment and many of them do not know the meaning of US timezone abbreviations.
If you insist in dual time presentation please use something like the W3C datetime profile, with local timezone defined as an UTC hour offset (see last two examples of http://www.w3.org/TR/NOTE-datetime)
On Friday 29 June 2007 10:30 am, Matej Cepl wrote:
On Fri, 29 Jun 2007 10:09:55 -0400, Jeff MacDonald scripst:
+1 to use ISO-style, but please list UTC *and* local time for the outage.
What local time? Why should EDT matter to me or other people living somewhere else than on the eastern shore of the United States (and maybe in Brazil, I am not sure)?
good point.. local time should be the time local to the physical location of the equipment in question.
Matěj
regards,
On Fri, 2007-06-29 at 10:53 -0400, Jeff MacDonald wrote:
On Friday 29 June 2007 10:30 am, Matej Cepl wrote:
On Fri, 29 Jun 2007 10:09:55 -0400, Jeff MacDonald scripst:
+1 to use ISO-style, but please list UTC *and* local time for the outage.
What local time? Why should EDT matter to me or other people living somewhere else than on the eastern shore of the United States (and maybe in Brazil, I am not sure)?
good point.. local time should be the time local to the physical location of the equipment in question.
And what good does that do? I think just UTC should be fine.
josh
The 'international-time' package in Fedora can be used to copy and paste local and UTC times of specific events (such as meetings, outages, etc) into emails.
Tim. */
On Fri, 2007-06-29 at 09:55 -0500, Josh Boyer wrote:
On Fri, 2007-06-29 at 10:53 -0400, Jeff MacDonald wrote:
good point.. local time should be the time local to the physical location of the equipment in question.
And what good does that do? I think just UTC should be fine.
It can have some use for the relatively few admins who were not present for the event but are subsequently working on the box to further develop what happened. Log files are generally written out in localtime.
For the rest of the population receiving the email... not so much.
Note: It might be a better idea to turn "localtime" into UTC on the equipment instead.... -Toshio
On Fri, 2007-06-29 at 16:30 +0200, Matej Cepl wrote:
On Fri, 29 Jun 2007 10:09:55 -0400, Jeff MacDonald scripst:
+1 to use ISO-style, but please list UTC *and* local time for the outage.
What local time? Why should EDT matter to me or other people living somewhere else than on the eastern shore of the United States (and maybe in Brazil, I am not sure)?
Speaking as someone from the midwestern United States, I am very used to dealing with PST and EST, as so many technological things are based on the east and west coasts. Online services, cable TV stations and whatnot. They're only an hour/two hours off from my local time anyway. Whereas when I see a UTC date, I just go "Huh? Is that today or tomorrow?". But then I've never been any good with numbers...
Seems to me, just posting both local time and UTC should cover both sides of the globe reasonably well.
On Monday 02 July 2007 4:48 am, Callum Lerwick wrote:
Seems to me, just posting both local time and UTC should cover both sides of the globe reasonably well.
+1
regards,