Feeds showing up unread despite being read seconds ago.

I know this doesn’t help, but the problem still exists with about 2 weeks of read story data, which means that any previously stuck stories will take 2 iterations of you reading them (one to unstuck, one to read) before they work. I know it’s less than ideal, but it’s a big system and change is becoming harder to make to a live database.

This will quickly catch up and within a week there should be no more. I’ll be actively watching what happens, so rest assured that I’ll keep working at it, even if you happen to find old stories that are still incorrectly marked as unread. New stories should not have this problem.

Ok, that’s fine. Knowing that it is not going to repeat is enough for me to go through this cleanup. Am I right to understand that the same feed should not show up with problems after I go through the two steps?

And thanks for the heads up!

Looks good so far, Samuel. Thanks! As expected, there were a few feeds where I had to run through multiple times but they have not acted up since. Yay.

I’m experiencing this (or something similar anyway). It’s just one feed, and one article in that feed. I can mark it as unread and the count goes to zero, but any time a legitimately new item appears in that feed, this old article shows up again as unread. It’s “NEW PRODUCT – 12′′ eTape Liquid Level Sensor + extras” on www.adafruit.com/blog/feed/. Been happening since Monday.

I just realized that I was still using a FeedRinse’d version of this feed, so that could be the cause. I’m switching to the regular feed and will report back if the issue persists.

Seems like the problem has cropped up again. 5 feeds keep on showing a few articles as unread even though I have read them. 5 times over.

Romit, I’m in your account trying to find stories that are sticking but I can’t find any. Are they in particular feeds? Has it cleared up now? I’m thinking it might have something to do with timing offsets between my servers, which might account for the shift between working and non-working in a matter of an hour.

It may have cleared. Some examples: All About Microsoft, Official Google Blog, Geekwire, Paidcontent, etc.

Just noticed - in my “Engadget” feed (which I have “reset” per your note earlier, many times), this story has become unread twice after I have marked it as read: http://www.engadget.com/2011/10/18/mo…

same thing with Geekwire feed, where the first article keeps popping up as unread.

Here’s the relevant logs:

[app01] [Oct 18 18:43:00] —> [IE] [rpmehta*] Read story in feed: Engadget
[app01] [Oct 18 18:43:00] —> [IE] [rpmehta*] Marked story as read: Duplicate Story -> http://www.engadget.com/2011/10/18/mo…
[app01] [Oct 18 18:43:00] —> [IE] [rpmehta*] Read now date: 2011-10-18 18:43:00.396396, story_date: 2011-10-18 18:20:00, story_id: http://www.engadget.com/2011/10/18/mo….
[app01] [Oct 18 18:43:00] —> [IE] [rpmehta*] Subscription mark_read_date: 2011-10-18 18:20:01, oldest_unread_story_date: 2011-10-18 18:34:00
[Oct 18 18:43:00] —> [IE] [rpmehta*] Original read date: 2011-10-18 18:37:42.317000, story id: http://www.engadget.com/2011/10/18/mo…, story.id: 4e9dc48782108873b4000001

It shows that you read the story 6 minutes earlier (and proof that it was the same story). The problem is that when the code grabs a list of read stories, this original story is not showing up, even though it’s part of the original list. It clearly has something to do with the fetching of read stories in these lines in the code: https://github.com/samuelclay/NewsBlu…. Somehow the story id is not being found.

Maybe I should hash the story id so it’s easier to detect. This might be a worthwhile change I can make this weekend.

and paidcontent too. sigh. please ping me when it is fixed. i am to OCD-ed to keep up with this any longer :frowning:

I noticed you’re using IE. Have you tried marking the story as read with Chrome? When I log in to your account I’m able to mark all of those stories as read. I’m wondering if IE is doing something odd to the story ids.

ok, phew :slight_smile: Can you let me know when I can check next? I don’t want to be in this whack-a-mole game :frowning:

I am going to use Safari for the next few days and see if there is a difference in behavior. Just cleaned up my current backlog.

…and, I see an old article on winrumors.com showed up as unread in Safari as well: http://www.winrumors.com/htc-windows-…

I checked your account again. It looks like everything is working fine under Safari. Is that the case? If it’s IE-specific, I’d be more than happy to look at it under IE, but I just want to be sure it is an IE bug.

Yes, almost confirmed that it is an IE-related bug. With Safari, no issues with dupes. I reset IE to factory settings and started all over again, and the issue showed up immediately.

I do all my development in Safari, and occasionally with Firefox. This means that Chrome, Safari, and FF are fully supported (as well as the iPhone app). I check IE maybe once a month, and before IE9 I certainly didn’t support it. I’ll see what I can do about IE, but it’s less than 4% of my visitors. Priorities, man.

Also, this issue was compounded by the fact that there were actually two issues – one I fixed a week ago (and seems to be helping a few other users as well) and this IE bug. I’m going to keep my eye on it.