The voice.suttacentral.net server is no longer working as expected. This problem was first reported by @Gillian (). Attempts to restore the server have failed and are still in progress.
If you need Voice functionality, please see EBT-Vue3, which is the newer version of Voice. EBT-Vue3 is still in development and may not have all the Voice functionality at this time. For example, EBT-Vue3 does not have audio downloads yet.
Our immediate priority is to continue working on the existing Voice server so that we can conduct a proper transition to EBT-Vue3.
Sorry for all the disruption! Ayya @Sabbamitta and I will keep you posted on latest developments here.
We’ve replaced the Voice website with a simple outage page that redirects users to EBT-Vue3. This will allow us to continue our investigation independently while allowing users to listen to the suttas. EBT-Vue3 does not support downloads, but we do have a small, fragile server that we can let you use. If you need downloads, let us know and we can provide you the URL for Voice downloads.
Thank you @karl_lew for all your work to get the problem fixed so fast!
To avoid misunderstandings, we should note that a new server doesn’t mean newer software. It’s the same old software, the new server is just a copy of the old one. Software updates are no longer possible for Voice, and the future will lie in our new application EBT-Vue3.
However, and this comes as a surprise even for Karl and me, Karl has been able to update the content, so Voice is, after a long time, again up-to-date with changes and additions of translations. And we are confident that we can continue this process of updating content regularly.
So in the end of the day, the new server is even better than what we had before!
For the curious, the root cause of failure was that the disk filled up with audio recordings. When a computer runs out of disk space, it is REALLY bad since the disk is used to store everything immediately useful. We are caching audio recordings on the server as well as on Amazon S3 storage to reduce costs. For example, the Pali audio cache for Aditi just for MN alone was about 2 gigabytes of data. As you can see, the multiplicity of narrators, languages and suttas creates a surprisingly large abundance of data.
Efforts to repair the server proved unsuccessful. Even after removing 12GB of cached data, the server failed to behave as expected. So we had to create a new server for Voice. Servers in the cloud are, from our perspective, all virtual. We didn’t need to purchase new hardware, we simply needed to allocate a new virtual server and configure for use. Given the age of Voice software, we were a little worried about being able to create a new server. Fortunately, the older software is still functional and we were able to create two virtual servers. The second virtual server we will hold in reserve so that we can bring online in the event of future outages.
As Ayya Sabbamitta mentions, we are moving Voice functionality to our new application platform EBT-Vue3. Voice is built on Vue2, which is deprecated. We rewrote EBT-Vue3 on top of Vue3, which is drastically different from Vue2, so the transition has been difficult. Because of that difficulty we took some time to redesign and improve Voice functionality to enable a nimbler “play this segment now” user interaction. We’re not done yet with EBT-Vue3, but it works well enough to be useful.
Unfortunately I get the result below, I tried the search 3 times and it consistently counted 1 less and gives MIL2, so I searched for MIL4 and it returned MIL4 OK.
Does it work for texts like the Milindapanha? I’d never expected that! I only ever searched for texts that have been segmented … I am surprised!
But I remember that I noticed at some other occasion a while ago that there were some mismatches with the numbering in Mil, though I don’t remember the details.
In any case, if it’s a Voice bug, we don’t currently have the possibility to fix anything inside of Voice.
Wow. This is a new bug. I never heard of MIL3.1.1, so we never tested for it. Thanks for discovering this. That sutta has three (3) numbering parts. I’ll look into this. I see only Pali files, but that should be supported.
). Attempts to restore the server have failed and are still in progress.
I’d never expected that! I only ever searched for texts that have been segmented … I am surprised!
