Upcoming Maintenance on Intel Map

245

Comments

  • Gut feel is that it was slightly slower than I expected - but that could just be me.

    I've noticed some requests taking a little longer. The longest one I saw was 24s for a getEntities call. But I'll take a slightly slower, completely accurate map, over an inaccurate fast map.

  • I went through another long load's logs, and saw 3 things.

    1. A call that completed successfully after 36 seconds.
    2. A call that failed with a 500 error after 51 seconds.
    3. A call that failed with a 500 error after 1.3 minutes.
  • Accurate but slow is better, I always have 'debug data tiles' enabled so I know (more or less) when it has finished.

  • That's great, but does it mean we're going to have finally new features on the Intel Map?

    *Looking forward for the answer, hopeless*

  • ofer2ofer2 ✭✭✭✭✭

    @Perringaiden thanks for the info! I think I've said this previously, but 500 errors are a separate issue that is being worked on, but its more complicated and will likely take a bit longer to fix. With the changes in the last month or so, the dashboard should be running much faster than it was previously. The new servers should currently be trading off a bit of that speed due to incrementally moving to a new environment instead of all at once. So, hopefully there should be more perf improvements coming up (with no loss in correctness).

  • ofer2ofer2 ✭✭✭✭✭

    No promises, but what sort of feature would you want the most?

  • Ability to bookmark or favourite portals would be very helpful, that and S2 cell overlays are the only two reasons I use IITC.

  • edited January 2020

    Ok so an update 8 hours after the last load of my LA reference map.

    • 43015 portals loaded. Which is +5 from the earlier load today, an increase that was expected based on new portals coming online.
    • Time taken was 31.8 minutes. This is slightly slower than it used to take in November (usually around 25 minutes)
    • There were 3x Status 500 Errors, all due to long getEntities calls timing out at 59s, 1.4 mins and 1.7 minutes respectively.
    • All up there were 2712 getEntities calls, so 0.13% failure rate.

    I would say that whatever change made today was a success in terms of accuracy. As you mentioned, @ofer2, the transition will be causing some lag which will go away in time, so hopefully it'll return to the same speed it had pre-December.

  • On initial load of this close view reference, I could not find any portals missing at all.

  • ToxoplasmollyToxoplasmolly ✭✭✭✭✭

    Codename: @Toxoplasmolly

    Location: https://intel.ingress.com/intel?ll=43.073603,-89.40556&z=17&pll=43.073603,-89.40556

    Type of Issue: Other: Destroyed links and fields still being shown

    Platform: Desktop (Chrome, macOS)

    Approximate time of incident: 2020-01-08 09:20 CST (UTC-6)

    Screenshot: 

    Other notes: I originally loaded this view in IITC. Then I loaded it in stock intel for the purposes of this screenshot. COMM entries indicate that those links were destroyed at 09:07, over ten minutes before the screenshot was taken and thus beyond an acceptable timeframe for “intel lag.”

  • EvilSuperHerosEvilSuperHeros ✭✭✭✭✭

    Loading 43k portals at once over the entire LA area? Do people really need to do that all that frequently or at all? Seems like it would take up quite a bit of resources. I can barely get the same 150 odd portals to load up over the local park and you are pulling down 43k ?

  • It's not a constant thing, but it's been a good indicator of the difficulties of the intel map, because at that scale, even if one area loads where it didn't the day before, the aggregate always shows the issues.

  • tehstonetehstone ✭✭✭
    edited January 2020

    Ability to see all images a portal has.

    Name of the agent who original created the portal.

    Date portal was created.

  • The number one reason people use IITC is its refresh rate and Bookmark functionality. For ops involving large fields Intel is useless. Bookmarks alone makes IITC the standard.

    Also, Intel will only reload 40 links at a time. So if you try and bookmark a field plan even on a small scale it will only show the first 40 links drawn, which is unacceptable. I've drawn field plans with over 120 links!!! The first time I drew a field plan on stock Intel and it didn't load half my plan, was the last straw for me.


    So for me and everyone else. IITC style bookmarking and unlimited drawn links saved.


  • Actually one of my major reasons for using IITC is hiding fields and links, or hiding portals, to focus on one or the other. Stock Intel can 'sort of' hide portals, but not fields or links.

  • FraggerFragger ✭✭
    edited January 2020

    At the start of a new cycle/end of the last, show the data for checkpoint 35 of the last cycle instead of nothing for the current, since cp 1 has yet to happen.

  • Why would want to hide portals? I'd prefer highlighting the ones you want to focus on (filtered by level, favorites, etc.). You get a more clear picture of the field of play. A lot of my field planning is determined by ease of clearing. I've thrown over 2 million MU by myself because I found the lanes lightly obstructed enough that I could attempt the rest of the clearing by myself even though the drive time was cutting it close. All in all, just over a 100 miles driven over 30 layers on a 3 spine field pattern in two and a half hrs. Yopu want to see every portal and link in as real a time as possible, comm activity too! Hiding portals and links is like the last thing you should ever want except to just see what your field is going to look like alone and by itself, which I guess is useful...maybe?

  • @ofer2 if you're entertaining any requests for the Intel page, it would be really nice to have a way to request HISTORIC scores for the septicycles. If not the full history (which might be a bit much, but it'd be nice) then certainly to display the final score (who won/lost) from the previous cycle.

    Currently, when checkpoint 35 hits, the scores go immediately to show 0-0, rather than showing the final result of the septicycle. On a close cycle, you might not even know who won...

    Anyway, just sharing a wish list. Definitely think first order of business is getting the map to show an accurate game state. And thank you.

    (Edited because I was on the wrong thread)

  • edited January 2020

    Not everything is about fielding. Sometimes you want to hide everything else so you can decide where to farm.


    And I was talking about a way to get specific information from a larger areas without having to load then discard everything else.

  • If you are looking for P8, then it is handy to hide P0-P7.

    I've a plugin (modified from less-clutter) that displays all portals fully transparant (ie invisible), however highlights still show - makes spotting uniques easier.

    I sometimes hide resistance+enlightenment+P0 - just to get a nice pannable map with no screen space wasted on borders and the ability to flip between mapping databases!

  • ofer2ofer2 ✭✭✭✭✭

    @Perringaiden we're going to be doing another test today at 10:30 ish. Please let me know if any issues pop up.

  • edited January 2020

    @NianticOfer I cannot post a screenshot because of the stupid Vanilla Forums filter that's enabled, but we're getting 502 errors again. My stock reference from 14:35 PST is missing two large chunks and had 1x 502 error, and the LA reference had 19x 502 errors during load.

    That said, requests are completing much faster than they were yesterday, with the majority under 2 seconds instead of in the tens of seconds.

    So we've traded off accuracy for speed at this point.

  • ofer2ofer2 ✭✭✭✭✭

    @Perringaiden As I've said previously, the 502s are a separate issue. That issue is mostly a contention issue: i.e. the more the same resource is requested at a time, the more errors you will see. This means that the error rate is largely random. I will let you know when we have a fix for that. To be fair though, we probably made contention worse by making the servers run faster....

  • ofer2ofer2 ✭✭✭✭✭

    Alright, we're turning the new servers off due to a separate issue. Will update you all when they're back online.

  • What? Motoz3 Android

    Why?



    Dear Omnis-Digital Inc., 

    What does this mean?


    Error: disallowed_useragent

    Google can’t sign you in safely inside this app. You can use Google sign-in by visiting this app’s website in a browser like Safari or Chrome.

    Learn more

    Request Details

    That’s all we know.

Sign In or Register to comment.