Site Stats for May 25, 2006

[Howard Beasley]

The Site Stats for my blog for May 25, 2006 are identical to the stats for the previous day, May 24, 2006. Since this is highly improbable (all four of the Summary Stats are identical), I suspect there has been an error in posting of these numbers. Anyone else with this problem or any insights as to its cause?

[john]

It was an error on our part and will be corrected.

[Howard Beasley]

Many thanks. (Great service by the way.)

[gristgal]

Ohmigosh, I didn't even notice! But you're right.

Howard, you have my undying gratitude.

John, any guesstimates on the time frame?
_________________
"Real knowledge is to know the extent of one's ignorance."
Confucius

[Howard Beasley]

The stats on my blog were corrected by Saturday afternoon and I remain a happy camper. Now I'm back to waiting for Florida Power & Light to remove a broken pole that has been in our back yard since Katrina. Each of the several times that I've called they have promised to come get it in three days. Blogharbor has noticeably better customer service.

[john]

John, any guesstimates on the time frame?

Site stats were revised the same day, 5/26. Though in anticipation of your next question about server logs, I checked and they were not revised. I've put in a note on this and we'll see what we can do about getting them in there.

The stats on my blog were corrected by Saturday afternoon and I remain a happy camper. Now I'm back to waiting for Florida Power & Light to remove a broken pole that has been in our back yard since Katrina. Each of the several times that I've called they have promised to come get it in three days. Blogharbor has noticeably better customer service.

There's a new opportunity for a catch phrase there: BlogHarbor - better than your local utility monopolies!

[gristgal]

Mine appear to be fixed, but the new ones look weird. I'll look at them and see if I can figure out why. The day's stats - pages viewed and served - look consistent with subsequent days, and double Tuesday's, so I'll hafta see if there's some other way that there could be that much activity, and still very few server requests.

Any ideas, John?

I hate to ask it, but could you take a look, too, please? ISTM that the various measures should be consistent with each other.

[gristgal]

Whatever you did, thanks! 11.9 kb was the incongruously small file. The new one, at 161 kb, is much more plausible.

Is it too late for you to take a look at the stats for the 23rd? I think in the relief just to have data for those days I didn't think about the size of the log files. For that day, I show 351 requests, but only a 55 kb logfile. I guess it just wasn't small enough to register as being fishy, with everything else that was going on. Yesterday, there were only 166 requests, so I guess that a logfile of 43 kb is about right?

Sorry to be so much trouble!

[john]

Is it too late for you to take a look at the stats for the 23rd?

I did look at it the other day and I did not see anything particularly unusual about it...

[gristgal]

On the 22nd, there were 325 total reqs (page views), and a logfile of 117k
On the 23rd, there were 394 reqs, and a logfile of 55k
On the 24th, there were 311 reqs, and a logfile of 139k
On the 25th, there were 389 reqs, and a logfile of 161k
On the 26th, there were 376 reqs, and a logfile of 216k
On the 27th, there were 366 reqs, and a logfile of 247k
On the 28th, there were 434 reqs, and a logfile of 266K
On the 29th, there were 186 reqs, and a logfile of 120K
On the 30th, there were 374 reqs, and a logfile of 70K
On the 31st, there were 138 reqs, and a logfile of 70k

I could give more, but that should be more than enough. ISTM that the size of the logfile for the 23rd is notably inconsistent with the number of requests.

Two more problems have cropped up since. The logfile for the 30th is again inconsistently small (as you can see above). And the logfile for the 31st appears, upon inspection of the first record, to be a duplicate of the 30th. The dates read 30/May/2006, and the first request is identical. I didn't read any farther.

For all I know, you're still working at making the new equipment configuration work properly together. If that's the case, I apologize for adding to your woes, but AFAICT these are all real glitches.

If you wish it, I will go into those two files (23rd & 30th), separate and count the requests for each of the days I believe have missing data. I have looked at a sample of daily logs, both small and large (for my blog, anyway; I'm sure there are many others you host whose logs are vastly larger because they have many times the number of readers I do. I'm just small fry.).

None of the logs I've glanced over appears anomalous in any particular except size. In general, days with more than 300 page views have logs that are greater than 100k, usually much greater. I realize there can be some variance in the length of each command, and it's possible that the majority of requests processed on those days came from machines with short ID strings, sent through servers that also have short IDs. However, I believe that for that to happen with several hundred reqs and more than a very few readers making the reqs, the probability is low. Because the number of hosts on each of those days was lower than average, I am not saying it's impossible. If the number of hosts were really small on either day, I might have a lot of trouble accepting it, but I would. But they're not that small. I haven't had a day with that few visitors in a while.

[john]

I could give more, but that should be more than enough. ISTM that the size of the logfile for the 23rd is notably inconsistent with the number of requests.

The size of your log file won't necessarily correspond to the number of page views for a variety of reasons... A few of which might be that the log file contains more than just requests for page views (trackbacks and images) so let's say someone was sourcing an image file or you got a lot of trackback spam for example, your log file might be significantly larger

That being said, I just talked to our technicicans and they noted that they are still having a problem scraping all the logs from all the different servers which respond to access for your blog pages (there are 2 different arrays of servers, and your blog page might be served from a server in one array one minute then a different server in a different array the next) so the logs are not containing all of the data. They believe they have found the source of the issue, and a fix is undergoing quality assurance testing.

For all I know, you're still working at making the new equipment configuration work properly together. If that's the case, I apologize for adding to your woes, but AFAICT these are all real glitches.

We are in fact aware of the problem and working to resolve it. No need to apologize, you're not adding to our woes. Our woes are our own.

Sorry for the problems with log files, and for our delay in resolving them.

[gristgal]

[john]

I've been thinking that there seem to have been an awful lot of problems (all the new topics on this or that problem that none of us seemed to be seeing B.U. (Before Upgrade))

Can't agree with you there... Yes, we have had some problems, primarily with the stats processing, since the caching server installation. Outside of posts related to that issue, a quick review of the questions posted here in the last few weeks looks to me to be the typical collection of How do I, Is it possible to, and Why does it work that way type of questions...

Of course, I'm making the assumption that the daily stats summaries are correct. Those numbers are accumulated by a different process, aren't they?

Yes, they are.