Jump to content

prupdated

Members
  • Posts

    150
  • Joined

 Content Type 

Profiles

Downloads

IPS4 Providers

Release Notes

IPS4 Guides

IPS4 Developer Documentation

Invision Community Blog

Forums

Store

Everything posted by prupdated

  1. Bing seems happy with the way things are: In Google, the site has practically been buried.
  2. Random avatars I am back to using for new members work out like this ( selects from about 200 ):
  3. That would be good if it were built in to IPS. Right now it's pretty simple on nginx, with a few image_filter lines. So if you want a 250x250 (or whatever size) version of /image.jpg in the URL, you just call /250x250/image.jpg instead. Something that simple would be good to have to keep image sizes from being too large. Although, you have to be careful not to resize the same images in too many places. Then the client doesn't have the cache of all the sizes so it can be counter productive.
  4. I'd say that's some valuable feedback then is to at least generate a thumbnail image for these. That explains why earlier I saw blank thumbnail entries in the database for all of these which is how I wound up finding this topic. I used to use a random avatar assignment script to do all this. I turned it off and went with this new method. But, I didn't realize this was going on. That might be another idea or feedback, is to allow a pool of random avatars or something else be selected instead of letters. In any case, the letters are now off for me. I'm back to giving new members random avatars (with thumbnails) and also running a script now to revert back the 10,000 members who got these big letters.
  5. So there is no thumbnail being created for these avatars?
  6. My sitemap has everything in it. It's updated daily. Still I got slammed pretty good. Years ago, I probably had 90% of my sitemap indexed. Now it's less than 10%. Nothing much has changed with the actual type of content. I've had ssl for years. And redirects from that have long since cleared. So Google basically says most of my site isn't worthy of it's index. In the old days, it would be the other way around - 900k submitted and 850k indexed. I don't think a special formula of sitemaps and meta tags are the ultimate fix here. It appears to just be Google deciding what's important. The burden might just fall on the webmaster doing something their content versus Google's algorithm. Here's a pretty decent article going into the Fred update with some ideas on what to do about it: 3 Examples of Impact From the March 7, 2017 Google Algorithm Update (AKA Fred) https://www.gsqi.com/marketing-blog/march-7-2017-google-algorithm-update-fred/ Also see the Goole Quality Rater Guidelines: https://static.googleusercontent.com/media/www.google.com/en//insidesearch/howsearchworks/assets/searchqualityevaluatorguidelines.pdf
  7. I think it’s just Google doing it’s ‘thing’. I’m pretty sure this has little to do with sitemaps. I process all my sitemaps each night with a simple in-house script for my 500k or so topics. It only takes a few minutes to run. I have been doing it this way for years. All the links show up in Google as being in the sitemap. But about a year ago, Google started dropping them from the index by the hundreds of thousands. Also my home page key words used to be in the top 3 at Google, now they've been lowered into obscurity. It’s not blacklisted. Some links are still there but just not ranking. Yet, if I post something on YouTube or set up an empty store on a popular ecommerce site, all of a sudden those links are up at the top of some search results. So, it may just be Google on yet another new algorithm deciding who gets to see what.
  8. It could just be some bad email addresses in your user database. When you get bounces like these, as well as unsubscribes and complaints, you need to process that feedback and stop sending to them. The forum software doesn’t do that natively. It will just keep sending to them. You’ll need to either do it manually, program it yourself, or use an email service provider. To help not sending as many bounces, you can check your forum settings so that by default members only receive one notification until they return to the forum. Also review the email notifications that get sent out by default. There is also a bouncer application to help with this in the marketplace here. It links the bounce and other feedback from the email provider into your forum member database. Then it stops emailing and prompts them to update their email.
  9. Great ideas so far. Maybe you can write one? Many people here would surely love to see it. It doesn’t seem too complex. It almost seems like this should be built in. So far what I did seems to be working though in trimming off about half a million private messages from over 3 years ago. I can’t really wait another several years because it’s a struggle sometimes to keep the board running smoothly. The problem is with bloated tables like this, notifications, and search, some forums such as mine (several million posts) can really start bogging down. And I can’t really justify throwing more hardware at it. Once these tables are trimmed up and reoptimized, along with severely limiting search and notification history, and sticking mostly to core functionality, the site runs at about half the cpu / io or less than it would otherwise.
  10. Write about a 20 line php script to do the sql. There are three sql message tables that interact with each other. Do a select on one of the tables based on your criteria. Then loop through the results array and remove entries from all three tables with the matching private message topic id. Study the tables a little and understand what the fields and mappings are. It should take about an hour or two to write. Obviously, give members some advance notice of this, do a backup, and test with manual sql lines, debug output lines and small limits before operating on live data. Of course, if there's something else I'm missing that's linked in another table somewhere I don't know. I've already done it. It seems to work. You'll need to write it on your own though. I don't mind the risk on my own board. I have mine set for 36 months retention from the time of the last post in a PM topic except for the bot account which is 3 months.
  11. I needed the same thing. I wanted closed, but didn't want to manually approve the joins. There was a simple edit I made to one of the club PHP files to let this happen (it should probably be a built in feature to let clubs decide independently between access and join approval process). Funny thing also about the word closed. I renamed 'Closed' to 'Participant' in the language settings. I wasn't sure if it would confuse people seeing a closed group assuming that well, it was closed and they couldn't join it. Lastly, I noticed only a trickle of people joined the clubs even with auto accept of invites and front page and other club blocks giving them more attention. So, I wrote a script to invite members to certain groups at a rate of about 7 per minute (uses perl mechanize to login as a leader and do an invite) . I also set email notifications on first globally for club invites. This worked. I went from maybe 10 joins per day to 200. This might be a good idea for a future API call or a plugin - it helped quite a bit. You may find the clubs load slowly if you get a lot of joins in a day (500+). I had to drop my global search time to 2 days for now to get them to load quickly. There is also a global club permission setting. I excluded guests completely from clubs per the nature of my forum.
  12. Hi. Overall it looks pretty good except that I have 70,000 members and the script times out if I publish (60 second php timeout). It works on smaller groups. I haven't tried but it might work if I don't publish but wait for cron to pick it up. Although, I'm concerned what could happen if it is already timing out on the front end publish. Also, I'm not sure how long it would take using sparkpost since I don't see a settable sending rate with sparkpost. SMTP has a rate option but I run into some problems with SMTP on spark post from time to time. Any help would be appreciated. Thanks.
  13. prupdated

    Easy Pages

    The 4.1.14 update was one of the most disruptive updates to my site in the last year or two. Immediately after updating I got tons of complains about members not being able to login and other issues. I have been running Easy Pages for quite a long time. I also experienced some of the same issues noted above. It didn't stop with the Easy Pages app either. There were many other problems - all of which worked fine before 4.1.14. At first, I was able to partially work around the new login url problem by doing some server based rewrites / redirects. Ultimately, more problems surfaced and I gave up and converted what Easy Pages I had to the regular Pages app. It seems over time, the Pages app has added much of the Easy Pages functionality to it anyway. This was still hard because many of the features I used it for I had to change or get rid of completely during the migration. After this, I had to uninstall Easy Pages. Then I also had to reset my Furls to defaults. Then recached the site. This fixed most of the problems. User logins were now working properly rather than retuning error pages. However there were still problems unrelated to Easy Pages. It seems after resetting Furls to defaults, there were a few native IPS calls still using the old format that were not reflected in the new default Furls. So I had to go into the source code and work around those. And second to last of my problems, a number of url formats I used for custom theme changes previously were now broken - such as calling a topic with just a dash after it and nothing else now required two dashes. And the last present of 4.1.14 was some type of IPS search index task update that caused my sql server to nearly lock up so bad it caused my website to timeout while it was running. After a few days, I was able to isolate this as the problem and just set my site to only index 3 days of data so it would calm down this problem. There is a thread about this problem elsewhere on the forum. So what went from a seemingly routine IPS incremental update, turned out to be the nightmare of the year for me. All told it probably took me a week to sort out the mess. I will certainly take all updates very seriously from now on with full backups done immediately before so I can immediate roll back and research when this type of thing happens again. I will also likely skip many of the updates since they can cause so many problems. As it is, it has become more common to find an increasing number of plugins and updates and template changes and urls breaking on each IPS update. I can see why the cost to develop modifications for IPS 4 has gone up. Their shelf life may be only as long as a few weeks until the next update. At the same time, the number of modifications and functionality / performance of the forum software itself are a mere fraction of what used to be available in the IPB 3 days. For crying out loud, to this day I still have to rely on using a Google search for indexing my forum rather than the built in high performance Sphinx searching of days gone by.
  14. prupdated

    Portal

    Same problem here. I was able to work around it for now by editing this file: /applications/portal/extensions/core/FrontNavigation/Portal.php changed line reading: class _Portal to class _Portal extends \IPS\core\FrontNavigation\FrontNavigationAbstract I assume it has to do with some menu issue in 4.1 like is shown in one of the patches to 4.1.7 that refers to this type of error. I'm not sure if this workaround causes any other problems, but this at least got me going on 4.1.7 and easier than restoring back to 4.1.6.1. If you haven't upgraded yet, I would hold off until it's fixed though.
  15. I believe its: pp_setting_count_comments needs to default to 1 in core_members I ran this once and also as part of a script on new members to change that and some other defaults. The plugin mentioned might be good to use once you change all the existing members. When I first upgraded, I didn't even realize this was defaulted to off and though my site was broken.
  16. I've worked in the server IT departments of some very large companies before. There's no way we would jump on a major release like this until probably 6-12 months at the earliest. There would be a vocal set of internal customers who would want it immediately for whatever new features it has. But as the 'server guys' we have to support it. Once we put it out, and it breaks, we own it now. There are also a hundred and one unintended consequences that nobody foresaw - we have support tools, customers have third party apps they use, etc. I don't think it's unreasonable at all to expect at least a several month delay before anything like this should even be on the radar. At the end of the day, the software needs to be stable and reliable.
  17. I don't understand the guide without an example or more details about it. I did get infinite scrolling to work relatively easily on my IPB4 site with this however: http://infiniteajaxscroll.com/
  18. With 3.5 million posts, the fastest way I found was to: 1) Use cron to run tasks on my server and 2) reduce search depth to 90 days. It took about a week total. #1 - using cron - was essential for it getting done anytime soon. If you haven't set up your server to run IPS jobs via cron, I'd highly recommend getting it figured out no matter what it takes - especially as your forum size is larger. The server needs to be kicking off the cron jobs each minute regardless of how much activity is on the forum. The manual method in my experience was slower than cron and not as consistent.
    Very nice. Good to see moods back.
  19. Not that I know of offhand. That probably requires editing the cometchat files. Maybe you could just put a banner or something showing non premium members what they're missing out on. Besides, for load purposes, I generally avoid showing the bar to anyone who can't actually use it.
  20. It reads the member table from IPB. The reads are needed to authenticate, get avatars, and friends. As far as I can tell, it only writes to its own tables that it creates for the actual chat functions.
  21. Probably same thing being discussed here (also see a mod. in that thread that may be able to deal with some of this. I haven't tried it.)
  22. I just do this in the IPB global template so it only loads scripts for certain groups and if they are not banned. So something like: {{if !member.temp_ban and in_array(member.member_group_id, array('3','4','6','7','8'))}} ....replace with your group numbers above in the array to have access and put scripts here that load the bar.... {{endif}}
    Very nice layout for a more visual look rather than the traditional text based layout.
×
×
  • Create New...