Hi, I'm testing your fix directly on production after a backup. I have to say that the admin page is still remarkably slow as always. To load the order page we need to wait one minute or more. The site is still unusable. I'll check if we have also gateway error when we'll translate products with automatic translation. I have to say that if you don't solve the extreme slowness about our site in few weeks, we're considering (me and the store owner) to delete wpml and request a refund. If I disable sitepress-multilingual-cms and woocommerce-multilingual the site speed up a lot (few seconds to load an admin page vs At least a minute). Consider that if we update or add a product we cannot access to the admin page (only that) for some minutes (I set timeout to 6 minutes) every refresh. In some case we have gateway error.
After about 8 month we had lots of issues like we are not capable to add products in bulk, we cannot update translation title in bulk (it's impossible), we have issue with stock (wrong orders) because it's difficult to update that. Orders are less then the half than before the translations because we are not able to manage the shop. That's frustrating...
Honestly it seems to me a little improvement not the solution. I'll post as soon as I can the difference with a video about the site with and without WPML. I can understand a little loss of performance when active but but currently the performance with WPML is in an order of magnitude that make the site unusable by the administration.
I tried to change server, to add cache with Redis and WP-Rocket, to disable plugins (but after I reactivate considering that there are not the problem).
I'm a developer but I also engaged other more expert than me, but no-one can handle the issue.
I also tried to understand the issue on my own and seems to me that the issue is the way you manage jobs in the DB, maybe our products number and languages are unusual for your plugin.
Thank you for sharing the above feedback, as we are currently investigating the ERR_CONNECTION_REFUSED issue you are experiencing when sending jobs in bulk through the Bulk Auto Translate feature, let me know your results with the fix my colleague shared and if that helps this issue.
Once this has been confirmed we can move forward with investigating any performance issues you are experiencing, for that I will split the performance investigation into its thread once ready to ensure each issue is addressed.
I tested a little more the and it seems better but not solved when I start the automatic translation. One time I faced a server error when I selected more then 20 products to translate. But generally when they are listed the site seems to translate products. I'm trying to duplicate the site to test the product automatic translation by settings, but I had some issue to resolve in the duplicated site...I'll try again next days.
Hi, I test the
define('WPML_ATE_MAX_PARALLEL_DOWNLOADS', 1);
in the production machine with automatic translation enable. It seems work well.
I also have big news. Since it's holiday I start to made a lot of tests and virtual machine as I did in other occasion. I upgrade my server machine (AWS Bitnami lighsail) with 16GB RAM and 4CPU. That's a lot for a machine that have only WordPress inside. In this occasion I start a new instance (and not duplicate with a snapshot) and I use the duplicator pro plugin (that you suggest me) to move the production machine.
Now the admin page seems work well. Before start this thread I move from SiteGround to OVH plesk, then to bitnami. I did it as a manual migration or with all in one migration and the issue remain.
Even if the issue is resolved for me I'm keeping alive for a while the old production site with the issue I had. It's there. hidden link
I hope as your developer team go inside and try to understand why I had this issue . To be more precise, the admin pages need a lot of time to load. Considering that now the domain is not under crawler, so they are unloaded. But it was with the main domain and the shop owner start to made changes in product pages or translation, sometimes we need more then 5 minutes to load a single admin page. For me you could test everything and change everything in hidden link to investigate. The user and password is in the old messages. If you want sftp and db location give me the appropriate form. This old server has 8GB RAM and 2CPU and the DB is MySQL and it's hosted in another AWS lightsail instance to have more resources. The DB have 1 Core, 1 GB.
Now the new AWS Lightsail WordPress server have 16GB RAM, 4CPU and have both, DB and WordPress. The CPU is always between 30% and 40%. I think that's a lot for a WordPress site.
Is it possible that with your plugin enable this is the normal state?
Thank you
Thank you for your response and happy new year! I would like to ensure that I fully understand.
You have now moved the site to a new environment and also using the code my colleague provided define('WPML_ATE_MAX_PARALLEL_DOWNLOADS', 1); you are no longer seeing a performance issue, or do you? (I understand the old environment still shows an issue, but the current live one? Also did the above code help in the previous environment at all? )
That's true. I moved the site in a new environment and in this new environment I have no longer seeing performance issue. In the new environment the database and the site are in the same instance and can handle 4core and 16GB ram. I notice that when I use the automatic translation I use sometimes more than 60% of CPU, that's a lot.
I use the code your colleague provide in both environment, the new that now is in this hidden link
domain and in the old that I keep alive for debut purpose that's hosted here hidden link
With that line of code I notice that I have no more error about gateway but transaction are a lot slow.
A couple of times happens in the new environment that I finish memory. In addiction I notice that the automatic translation didn't recognize all product to translate. I start to add manually at the moment. In case I'll open a new thread if it will happens with the new products. I suppose that in the previous environment I have so many issue that some data are wrong in the DB
Thank you for sharing the above information as it's great feedback to share back to the team. May I propose that as the code provided and the new environment are working OK now we mark this ticket as resolved and if more issues arise please open a new thread as this information will all be under your account for us to reference.
Ok, you can close. But I hope you'll investigate. That's not normal that only to have a multi-language Ecommerce I need such powerful machine. If I disable it the store need a lot less power. With WPML and its services active I consume up to 50 times the CPU I need for the normal usage without WPML active in the site
I have been discussing this with the developers and they said that you can go ahead and shut down the older instance of your site.
There is currently a dedicated team working on performance issues, and high resource use is one of the areas already being worked on, and they have enough information about the sources of the problems already.
You can close this thread in the expectation that upcoming WPML releases will continue to make incremental improvements to performance.
Feel free to open a new thread at any time you experience some specific issue you would like addressed.
Ok, thank you. I just delete the old instance. I'm facing some other issue now that I'm using it more deeper, but I think it is not relevant to this discussion or maybe I use something wrong so I'll keep to read documentation. In addiction I'll wait to translate everything, maybe some issue will be gone before open other thread.
Feel free to close this.
The topic ‘[Closed] The automatic translation has been stuck for a long time.’ is closed to new replies.
