For a project I was working on I need a queue which will be too large to hold in normal memory. I had been implementing it as a simple file where it would read the whole file take the first few (~100) lines, process them, then write back the updated queue with new instructions added and the old ones removed. However, since the queue became too large to hold in memory like this I need something different. Preferably someone can tell me a way to peel off just the first few lines of a file without having to look at the rest of the data. I had thought about using a database (MySQL probably with sorted insert timestamps) but I would heavily prefer to do it without for load and bandwidth reasons (several servers would have to all be sending and receiving a lot of data from the DB). The language I'm working in is PHP but really this question is more about unix files I suppose. Any help would be appreciated.
Sucking out the first line of a file is pretty trivial (fopen() followed by an fgets()). Re-writing the file to remove completed jobs would be very painful, especially if you've got multiple concurrent servers working off the same queue file.
One alternative would be to use a seperate file for each job. If you have some concurrency-safe method of generating an incrementing ID for these files, then it'd be a simple matter of picking out the file with the lowest id for the oldest job, and generating a new id for each new job. You'd have to figure out some file locking, though, to keep two+ servers grabbing the same file at the same time, however.
I had same problems while I was working on enqueue/fs transport. I failed to modify a small portion at the begging of the file without copying it to the memory and saving back. Instead, but that's possible to do that with the end of the file. You can read a portion and then truncate it. That's not really a queue but a stack. So if you rely on message ordering this would not be a solution. In my case, I lock the file when the file has been read from the file, the lock is released.
This is how you could write messages to a queue file:
<?php
$rawMessage = 'this your message to put to the queue as a string';
$queueFile = fopen('/path/to/queue/file', '+a');
// here it may add some spaces so the message length is multiples of modular.
// that make it easier to read messages from a file.
// lock file
$rawMessage = str_repeat(' ', 64 - (strlen($rawMessage) % 64)).$rawMessage;
fwrite($queueFile, $rawMessage);
// release lock
This is how you could read messages from a queue file:
<?php
$queueFile = fopen('/path/to/queue/file', '+c');
// lock file
$frame = readFrame($file, 1);
ftruncate($file, fstat($file)['size'] - strlen($frame));
rewind($file);
$rawMessage = substr(trim($frame), 1);
// release lock
function readFrame($file, $frameNumber)
{
$frameSize = 64;
$offset = $frameNumber * $frameSize;
fseek($file, -$offset, SEEK_END);
$frame = fread($file, $frameSize);
if ('' == $frame) {
return '';
}
if (false !== strpos($frame, '|{')) {
return $frame;
}
return readFrame($file, $frameNumber + 1).$frame;
}
For the locking I'd suggest using Symfony LockHandler or simply take enqueue/fs.
Related
I have a small web-page that delivers different content to a user based on a %3 (modulo 3) of a counter. The counter is read in from a file with php, at which point the counter is incremented and written back into the file over the old value.
I am trying to get an even distribution of the content in question which is why I have this in place.
I am worried that if two users access the page at a similar time then they could either both be served the same data or that one might fail to increment the counter since the other is currently writing to the file.
I am fairly new to web-dev so I am not sure how to approach this without mutex's. I considered having only one open and close and doing all of the operations inside of it but I am trying to minimize time where in which a user could fail to access the file. (hence why the read and write are in separate opens)
What would be the best way to implement a sort of mutual exclusion so that only one person will access the file at a time and create a queue for access if multiple overlapping requests for the file come in? The primary goal is to preserve the ratio of the content that is being shown to users which involves keeping the counter consistent.
The code is as follows :
<?php
session_start();
$counterName = "<path/to/file>";
$file = fopen($counterName, "r");
$currCount = fread($file, filesize($counterName));
fclose($file);
$newCount = $currCount + 1;
$file = fopen($counterName,'w');
if(fwrite($file, $newCount) === FALSE){
echo "Could not write to the file";
}
fclose($file);
?>
Just in case anyone finds themselves with the same issue, I was able to fix the problem by adding in
flock($fp, LOCK_EX | LOCK_NB) before writing into the file as per the documentation for php's flock function. I have read that it is not the most reliable, but for what I am doing it was enough.
Documentation here for convenience.
https://www.php.net/manual/en/function.flock.php
Have a file in a website. A PHP script modifies it like this:
$contents = file_get_contents("MyFile");
// ** Modify $contents **
// Now rewrite:
$file = fopen("MyFile","w+");
fwrite($file, $contents);
fclose($file);
The modification is pretty simple. It grabs the file's contents and adds a few lines. Then it overwrites the file.
I am aware that PHP has a function for appending contents to a file rather than overwriting it all over again. However, I want to keep using this method since I'll probably change the modification algorithm in the future (so appending may not be enough).
Anyway, I was testing this out, making like 100 requests. Each time I call the script, I add a new line to the file:
First call:
First!
Second call:
First!
Second!
Third call:
First!
Second!
Third!
Pretty cool. But then:
Fourth call:
Fourth!
Fifth call:
Fourth!
Fifth!
As you can see, the first, second and third lines simply disappeared.
I've determined that the problem isn't the contents string modification algorithm (I've tested it separately). Something is messed up either when reading or writing the file.
I think it is very likely that the issue is when the file's contents are read: if $contents, for some odd reason, is empty, then the behavior shown above makes sense.
I'm no expert with PHP, but perhaps the fact that I performed 100 calls almost simultaneously caused this issue. What if there are two processes, and one is writing the file while the other is reading it?
What is the recommended approach for this issue? How should I manage file modifications when several processes could be writing/reading the same file?
What you need to do is use flock() (file lock)
What I think is happening is your script is grabbing the file while the previous script is still writing to it. Since the file is still being written to, it doesn't exist at the moment when PHP grabs it, so php gets an empty string, and once the later processes is done it overwrites the previous file.
The solution is to have the script usleep() for a few milliseconds when the file is locked and then try again. Just be sure to put a limit on how many times your script can try.
NOTICE:
If another PHP script or application accesses the file, it may not necessarily use/check for file locks. This is because file locks are often seen as an optional extra, since in most cases they aren't needed.
So the issue is parallel accesses to the same file, while one is writing to the file another instance is reading before the file has been updated.
PHP luckily has a mechanisms for locking the file so no one can read from it until the lock is released and the file has been updated.
flock()
can be used and the documentation is here
You need to create a lock, so that any concurrent requests will have to wait their turn. This can be done using the flock() function. You will have to use fopen(), as opposed to file_get_contents(), but it should not be a problem:
$file = 'file.txt';
$fh = fopen($file, 'r+');
if (flock($fh, LOCK_EX)) { // Get an exclusive lock
$data = fread($fh, filesize($file)); // Get the contents of file
// Do something with data here...
ftruncate($fh, 0); // Empty the file
fwrite($fh, $newData); // Write new data to file
fclose($fh); // Close handle and release lock
} else {
die('Unable to get a lock on file: '.$file);
}
The problem
How to write data to start of file if I have not enough space to allocate it in RAM and I have not enough space to make it's copy on current FS partition? I.e. I have a file with 100Mb size, I have 30Mb memory limit in my PHP script (and it can not be adjusted in any way) and I have only 50Mb free on my current FS partition. I want to add 2-10 rows to file (it's definitely less than remaining 50Mb FS space)
Some background
I know about XY-problem and agree that it's true for this case. But to reconsider this case I'll need to change significant part of current application (actually, it went from previous team) and, may be, API of other applications that using this file.
My attempt
I have not found solution for this yet. My previous approach was - to use some network buffer (i.e. to connect to some external storage, such as MySQL, for example - it's located on another machine where there is enough space to write file's copy)
The question
So, is it possible to write data to file's start when I have not enough space to allocate it in RAM and have not enough space to create file's copy on FS? Is using network (external) storage the only solution?
Say you want to write 2K to the beginning of a file, your only real option is to:
open the file
read as much from the end of the file as you can fit into memory
write it back into the file 2K later than you started to read
continue with the previous block of data until you have shifted the entire content of the file 2K towards the end
write your 2K to the beginning
To visualize that:
|------------------------|
|-----------------XXXXXXX|
------>
|-------------------XXXXXXX|
|----------XXXXXXX---------|
------>
|------------XXXXXXX-------|
...repeat...
Note that this is a very unsafe operation which edits the file in place. If the process crashes, you're left with a file in an inconsistent state. If you don't have enough room on disk to duplicate a file you arguably shouldn't work with that file and expand your storage capacity first.
#deceze hint me great idea. So I've finished with:
function reverseFile($sIn, $sOut, $bRemoveSource=false)
{
$rFile = #fopen($sIn, 'a+');
$rTemp = #fopen($sOut,'a+');
if(!$rFile || !$rTemp)
{
return false;
}
$iPos = filesize($sIn)-1;
while($iPos>=0)
{
fseek($rFile, $iPos, SEEK_SET);
fwrite($rTemp, $tmp=fread($rFile, 1));
ftruncate($rFile, $iPos>0?$iPos:0);
clearstatcache();
$iPos--;
}
fclose($rFile);
fclose($rTemp);
if($bRemoveSource)
{
unlink($sIn);
}
return true;
}
function writeReverse($sFile, $sData, $sTemp=null)
{
if(!isset($sTemp))
{
$sTemp=$sFile.'.rev';
}
if(reverseFile($sFile, $sTemp, 1))
{
file_put_contents($sTemp, strrev($sData), FILE_APPEND);
return reverseFile($sTemp, $sFile, 1);
}
return false;
}
-it will be quite slow, but recoverable if process is interrupted (simply look to .rev file)
Thanks to all who participated in this.
I've tried code suggested by #AlmaDo, don't try it on real projects, or you will be burn in hell, it is VERY slow. (60MB file - processing 19minutes)
You can run shell script - https://stackoverflow.com/a/9533736/2064576 (processed 420ms, can not understand how much memory does it use)
Or try this php script - https://stackoverflow.com/a/16813550/2064576 (160ms, worked with memory_limit=3M, not worked with 2M)
Some background information
The files I would like to download is kept at the external server for a week, and a new XML file(10-50mb large) is created there every hour with a different name. I would like the large file to be downloaded to my server chunk by chunk in the background each time my website is loaded, perhaps 0.5mb each time, and then resume the download the next time someone else loads the website. This would require my site to have atleast 100 pageloads each hour to stay updated, so perhaps abit more of the file each time if possible. I have researched simpleXML, XMLreader, SAX parsing, but whatever I do, it seems it takes too long to parse the file directly, therefore I would like a different approach, namely downloading it like described above.
If I download a 30mb large XML file, I can parse it locally with XMLreader in 3 seconds(250k iterations) only, but when I try to do the same from the external server limiting it to 50k iterations, it uses 15secs to read that small part, so it would not be possible to parse it directly from that server it seems.
Possible solutions
I think it's best to use cURL. But then again, perhaps fopen(), fsockopen(), copy() or file_get_contents() are the way to go. I'm looking for advice on what functions to use to make this happen, or different solutions on how I can parse a 50mb external XML file into a mySQL database.
I suspect a Cron job every hour would be the best solution, but I am not sure how well that would be supported by webhosting companies, and I have no clue how to do something like that. But if thats the best solution, and the majority thinks so, I will have to do my research in that area too.
If a java applet/javascript running in the background would be a better solution, please point me in the right direction when it comes to functions/methods/libraries there aswell.
Summary
What's the best solution to downloading parts of a file in the
background, and resume the download each time my website is loaded
until its completed?
If the above solution would be moronic to even try, what
language/software would you use to achieve the same thing(download a large file every hour)?
Thanks in advance for all answers, and sorry for the long story/question.
Edit: I ended up using this solution to get the files with cron job scheduling a php script. It checks my folder for what files I already have, generates a list of the possible downloads for the last four days, then downloads the next XMLfile in line.
<?php
$date = new DateTime();
$current_time = $date->getTimestamp();
$four_days_ago = $current_time-345600;
echo 'Downloading: '."\n";
for ($i=$four_days_ago; $i<=$current_time; ) {
$date->setTimestamp($i);
if($date->format('H') !== '00') {
$temp_filename = $date->format('Y_m_d_H') ."_full.xml";
if(!glob($temp_filename)) {
$temp_url = 'http://www.external-site-example.com/'.$date->format('Y/m/d/H') .".xml";
echo $temp_filename.' --- '.$temp_url.'<br>'."\n";
break; // with a break here, this loop will only return the next file you should download
}
}
$i += 3600;
}
set_time_limit(300);
$Start = getTime();
$objInputStream = fopen($temp_url, "rb");
$objTempStream = fopen($temp_filename, "w+b");
stream_copy_to_stream($objInputStream, $objTempStream, (1024*200000));
$End = getTime();
echo '<br>It took '.number_format(($End - $Start),2).' secs to download "'.$temp_filename.'".';
function getTime() {
$a = explode (' ',microtime());
return(double) $a[0] + $a[1];
}
?>
edit2: I just wanted to inform you that there is a way to do what I asked, only it would'nt work in my case. With the amount of data I need the website would have to have 400+ visitors an hour for it to work properly. But with smaller amounts of data there are some options; http://www.google.no/search?q=poormanscron
You need to have a scheduled, offline task (e.g., cronjob). The solution you are pursuing is just plain wrong.
The simplest thing that could possibly work is a php script you run every hour (scheduled via cron, most likely) that downloads the file and processes it.
You could try fopen:
<?php
$handle = fopen("http://www.example.com/test.xml", "rb");
$contents = stream_get_contents($handle);
fclose($handle);
?>
What's the cleanest way in php to open a file, read the contents, and subsequently overwrite the file's contents with some output based on the original contents? Specifically, I'm trying to open a file populated with a list of items (separated by newlines), process/add items to the list, remove the oldest N entries from the list, and finally write the list back into the file.
fopen(<path>, 'a+')
flock(<handle>, LOCK_EX)
fread(<handle>, filesize(<path>))
// process contents and remove old entries
fwrite(<handle>, <contents>)
flock(<handle>, LOCK_UN)
fclose(<handle>)
Note that I need to lock the file with flock() in order to protect it across multiple page requests. Will the 'w+' flag when fopen()ing do the trick? The php manual states that it will truncate the file to zero length, so it seems that may prevent me from reading the file's current contents.
If the file isn't overly large (that is, you can be confident loading it won't blow PHP's memory limit), then the easiest way to go is to just read the entire file into a string (file_get_contents()), process the string, and write the result back to the file (file_put_contents()). This approach has two problems:
If the file is too large (say, tens or hundreds of megabytes), or the processing is memory-hungry, you're going to run out of memory (even more so when you have multiple instances of the thing running).
The operation is destructive; when the saving fails halfway through, you lose all your original data.
If any of these is a concern, plan B is to process the file and at the same time write to a temporary file; after successful completion, close both files, rename (or delete) the original file and then rename the temporary file to the original filename.
Read
$data = file_get_contents($filename);
Write
file_put_contents($filename, $data);
One solution is to use a separate lock file to control access.
This solution assumes that only your script, or scripts you have access to, will want to write to the file. This is because the scripts will need to know to check a separate file for access.
$file_lock = obtain_file_lock();
if ($file_lock) {
$old_information = file_get_contents('/path/to/main/file');
$new_information = update_information_somehow($old_information);
file_put_contents('/path/to/main/file', $new_information);
release_file_lock($file_lock);
}
function obtain_file_lock() {
$attempts = 10;
// There are probably better ways of dealing with waiting for a file
// lock but this shows the principle of dealing with the original
// question.
for ($ii = 0; $ii < $attempts; $ii++) {
$lock_file = fopen('/path/to/lock/file', 'r'); //only need read access
if (flock($lock_file, LOCK_EX)) {
return $lock_file;
} else {
//give time for other process to release lock
usleep(100000); //0.1 seconds
}
}
//This is only reached if all attempts fail.
//Error code here for dealing with that eventuality.
}
function release_file_lock($lock_file) {
flock($lock_file, LOCK_UN);
fclose($lock_file);
}
This should prevent a concurrently-running script reading old information and updating that, causing you to lose information that another script has updated after you read the file. It will allow only one instance of the script to read the file and then overwrite it with updated information.
While this hopefully answers the original question, it doesn't give a good solution to making sure all concurrent scripts have the ability to record their information eventually.