Function execution happening out of order - node.js - node.js

I have this function for deleting specific jars from a remote server:
deleteJars: function(appDir, version, callback) {
fs.readFile('/file/location', function(err, data) {
if(err) throw err;
var array = data.toString().split("\n");
for(i in array) {
if (array[i].indexOf('worker') > -1){
var ip = array[i].split(" ");
var ssh = new SSH2Utils();
var server = {
host: ip[0],
username: username,
password: password
};
var myfiles = ssh.exec(server, 'rm ' + appDir + '/' + version + '/jars/myjar*.jar', function(err,stdout,stderr, server, conn, response){
if(err) console.log('No jars to delete');
conn.end();
callback(response);
});
}
}
});
}
It gets called in my application with this:
runningService.deleteJars(appDir, version, function() {
});
Immediately after this I have a smilar call to a copyJars finction which copies new jar files to the same location and after that a job is run which uses trhe jars. My problem is that sometimes the copy is done before the delete so the new jars are copied to the folder and immediately deleted with the old ones. Have I done something wrong with my delete function that allows the application to continue to the next step before completing the delete?

{
deleteJars: function(appDir, version, callback) {
fs.readFile('/file/location', function (err, data) {
if (err) throw err;
var array = data.toString().split("\n");
const promises = [];
for (i in array) {
if (array[i].indexOf('worker') > -1) {
var ip = array[i].split(" ");
var ssh = new SSH2Utils();
var server = {
host: ip[0],
username: username,
password: password
};
promises.push(
new Promise((resolve, reject) => {
ssh.exec(server, 'rm ' + appDir + '/' + version + '/jars/myjar*.jar', function (err, stdout, stderr, server, conn, response) {
if (err) reject('No jars to delete');
conn.end();
resolve(response);
})
})
)
}
}
Promise.all(promises).then((results) => { // results will be in order
// use the results array
callback(null, results);
}).catch(err => {
console.log(err);
})
});
}
}
With callbacks, maintaining order might be a little tough - but with Promises - we have a simple API called Promise.all that you can use like above. To continue using callbacks, you can look into libraries like async that have ways to deal with this.

Without seeing how you have this worked out with your copyJars function, this seems like a classic synchronous vs asynchronous issue.
You are using an asynchronous function, fs.readFile, with a callback
. If I understand the basis of how Node.js works correctly, it hands the operation of finding and opening the file up and reading the contents to the OS, when that is done, the OS returns to node and says "here it is" and then node executes the callback with the file data. This means that while the OS is off finding the files you want to delete, node will continue executing code which seems to be your copyJars function. Depending on how quickly the OS returns to your node process with the required information, this may not happen in the order you expected.
Some solutions may be:
You can use fs.readFileSync. This will execute synchronously and halt execution of other code while this is being performed. deleteJars sounds like it's not a one time thing, so this may not be the most efficient solution.
You can implement promises or look into async/await.

Related

node.js svn update/commit synchronously?

I'm using svn-spawn library to update/commit files to svn. Problem is my app calls svn up/commit in a loop, and because of the async nature of the call, svn-up is called from the next iteration of the loop before the previous svn-up can finish.
How to handle this issue? Is there any way to prevent the next call from happening until the previous one is complete?
Figured out a way to do it using async module.
async.series can be used to execute async tasks in a serial fashion.
This is how I did it.
function commitFile(arg, callback) {
svnClient.getStatus(filePath, function(err, data) {
//...
svnClient.commit(['Commit msg', filePath], callback);
//...
});
}
var toCommit = [];
for (var i = 0, len = requests.length; i < len; i++) {
//Adding files to commit, async.apply enables adding arguments to the anonymous function
toCommit.push(async.apply(function(arg, cb) {
commitFile(arg, cb);
}, 'arg1'));
}
async.series(toCommit,function (err, result) {
console.log('Final callback');
if(err) {
console.log('error', err);
} else {
console.log('result of this run: ' + result);
}
});
async.series needs an array of functions which must call a callback once they are done. It uses the callback to determine that the current function in done executing and only then it will pick the next function to execute.

Async not working as expected

I have several files that I need to write to remote server. As you can see in my code, I'm reading those files inside async scope and writing them to remote server as they become available. Instead of all files, only the last file is successfully transfer to remote server. I put console output statement inside readfile callback function and check path variable value. It outputs the last filename three times. So I'm suspecting that the last readfile call is overwriting the previous ones. Shouldn't async wait for the callback function call before it moves onto next iteration?
var attachments = ['1.jpg','2.jpg','3.jpg'];
var request = require('request');
var option = {
url: "http://www.xxxxxxx.xxxx",
headers: {"content-type": "multipart/form-data"}
};
async.each(attachments, function(att, callback){
path = imagePath + "\\" + att;
fstat = fs.statSync(path);
fsize = fstat["size"];
fs.readFile(path, function(err, data){
if(err)
{
return callback(err);
}
else{
option.body = data;
request.post(option,function(err,response,body) {
if (err) {
return callback(err);
}
else{
console.log(option);
callback();
}
});
}
});
}, function(err){
if (err) {
console.log(err);
}
else
{
console.log("finish");
}
});
Shouldn't async wait for the callback function call before it moves onto next iteration?
No, you're describing what async.eachSeries does. async.each runs each all tasks in parallel.
Your problem is that here is only one option object here. You perform console.log(option) inside of a function asynchronously run by request.post. That callback function runs after every request.post has been run. Since you only have on option object, it will have the final body value it was assigned by the last fs.readFile callback.
The solution here is to have every each function call create its own option object:
async.each(attachments, function(att, callback){
var option = {
url: "http://www.xxxxxxx.xxxx",
headers: {"content-type": "multipart/form-data"}
};

Node - how to wait on async operations?

Sorry, just starting with node. This might be a very novice question.
Let's say I have some code which reads some files from a directory in the file system:
var fs = require('fs');
fs.readdir(__dirname + '/myfiles', function (err, files) {
if (err) throw err;
files.forEach(function (fileName) {
fs.readFile(__dirname + '/myfiles/' + fileName, function (err, data) {
if (err) throw err;
console.log('finished reading file ' + fileName + ': ' + data);
module.exports.files.push(data);
});
});
});
Note that all of this occurs asynchronously. Let's also say I have a Mocha test which executes this code:
describe('fileProvider', function () {
describe('#files', function () {
it.only('files array not empty', function () {
assert(fileProvider.files.length > 0, 'files.length is zero');
});
});
});
The mocha test runs before the files are finished being read. I know this because I see the console.log statement after I see the little dot that indicates a mocha test being run (at least I think that is what is being indicated). Also, if I surround the assert with a setTimeout, the assert passes.
How should I structure my code so that I can ensure the async file operations are completed? Note that this is not just a problem with testing - I need the files to be loaded fully before I can do real work in my app as well.
I don't think the right answer is to read files synchronously, because that will block the Node request / response loop, right?
Bonus question:
Even if I put the assert in a setTimeout with a 0 timeout value, the test still passes. Is this because just putting it in a setTimeout kicks it to the end of the processing chain or something so the filesystem work finishes first?
You can implement a complete callback after all files have been read.
exports.files = [];
exports.initialize = initialize;
function initialize(callback) {
var fs = require('fs');
fs.readdir(__dirname + '/myfiles', function (err, files) {
if (err) throw err;
files.forEach(function (fileName) {
fs.readFile(__dirname + '/myfiles/' + fileName, function (err, data) {
if (err) throw err;
console.log('finished reading file ' + fileName + ': ' + data);
exports.files.push(data);
if (exports.files.length == files.length) {
callback();
}
});
});
}
You can call the file operation method by doing something like:
var f = require('./files.js');
if (f.files.length < 1) {
console.log('initializing');
f.initialize(function () {
console.log('After: ' + f.files.length);
var another = require('./files.js');
console.log('Another module: ' + another.files.length);
});
}
EDIT: Since you want to only have to call this once, you could initialize it once when the application loads. According to Node.js documentation, modules are cached after the first time they are loaded. The two above examples have been edited as well.
To avoid being caught up in nested callbacks. You might want to use async's each that will allow you to do the tasks asynchronously in a non-blocking manner:
https://github.com/caolan/async#each
I think that's a good test, the same thing would happen in any app that used your module, i.e. it's code could be run before files is set. What you need to do is create a callback like #making3 suggests, or use promises. I haven't used mocha, but there's a section on ascynchronous calls. You could export the promise itself:
module.exports.getFiles = new Promise((resolve, reject) => {
datas = [];
fs.readdir(__dirname + '/myfiles', function (err, files) {
if (err) {
reject(err);
return;
}
files.forEach(function (fileName) {
fs.readFile(__dirname + '/myfiles/' + fileName, function (err, data) {
if (err) {
reject(err);
return;
}
console.log('finished reading file ' + fileName + ': ' + data);
datas.push(data);
if (datas.length == files.length) {
resolve(datas);
}
});
});
});
}
chai-as-promissed lets you work directly with promises using eventually, or you can use the callback passed to your test I think:
describe('fileProvider', function () {
describe('#files', function () {
it.only('files array not empty', function (done) {
fileProvider.getFiles.then(function(value) {
assert(value.length > 0, 'files.length is zero');
done();
}, function(err) {
done(err);
})
});
});
});

How to do for loop in javascript without blocking

I have folowing script
var email_list = ['email1#email.com', 'email2#email.com',....'email100#email.com'];
for(i=0;i<email_list.length;i++){
if(checkEmail(email_list[i])){
//do processing save in db and email to email addresses.
}
}
This code will be blocking in nodejs how to make this non blocking?
You can do this without blocking the event loop at all, by using a recursive loop. This way what you end up with is only launching one database worker per call, at a give time. Assuming the database work you were doing was asynchronous, your code didn't really block the event loop. But the foor loop still launched a bunch of workers simultaneously, which will tend to clog the event loop(not block it). And you are right in that it is blocking the event loop while your for loop is counting from 0, to whatever the size of your array is. The following does exactly the same thing, but you only launch one database worker at a time(good), and you never count from 0 to length. Each worker is popped off the list after the work on the current email is done, and your global event loop is left to process other things, not email_list.length database requests simultaneously.
var email_list = ['email1#email.com', 'email2#email.com', 'email100#email.com'];
function checkEmailList(emails, emailCallBack, completionCallback) {
var someDataCollectdOverAllEmails = '';
function checkEmailAsync(email) {
db.doSomeDBWorkAsync(email, function (data) {
someDataCollectdOverAllEmails += data;
if (email_list.length) {
checkEmail(email_list.pop()); //If there are still emails to be checked, check the next one ine line
} else {
completionCallback(someDataCollectdOverAllEmails);//IF not, call the completionCallBack
}
emailCallBack(data);
});
}
checkEmailAsync(emails.pop());
}
function logIndividualEmailData(data) {
console.log('Sningle Email: ' + data);
}
function logGlobalEmailData(data) {
console.log('All Email Data: ' + data);
}
checkEmailList(email_list, logIndividualEmailData, logGlobalEmailData);
Process.nextTick example
process.nextTick(function () {
'use strict';
console.log('printed second');
while (true);
});
process.nextTick(function () {
'use strict';
console.log('never printed');
});
console.log('printed first');
Note however that in the example below, despite the fact that loopForever will run forever, it still allows both of our files to be read out. If we just had while(true) it would of course block and not allow this and one of our files data would not be printed out.
var files = ['blah.js', 'file.js'];
for(var i = 0; i < files.length; i++) {
fs.readFile(files[i], function (err, data) {
console.log('File data' + data);
function loopForver(loop) {//asynchronously loop forever, pretty cool, but only useful for really specific situations!
process.nextTick(function () {
if(loop) {
console.log('looping');
loopForver(true);
}
});
}
loopForver(true);
});
}
If I need to do stuff after the emails all send, I use the async library (docs), which provides some useful functions for control flow.
You will still need to rewrite checkEmail(email) into checkEmail(email, callback) as #S.D. suggests. In checkEmail you will want to call callback after everything is completed. This probably means that you will nest callbacks, calling the second async thing (sending the email) only after the first (db query) has completed successfully.
I also suggest that you follow convention by using the first callback argument as an err parameter. If you callback(null) you are explicitly saying 'there was no error'. #S.D.'s solution suggests instead callback(ok) which is the opposite of convention.
Here is an example showing a couple nested asynchronous functions and the async library.
edit - use async.eachLimit instead of async.each so you don't execute all 100 calls simultaneously
(function main(){
var emails = ["a#b", "c#d"];
var async = require('async');
async.eachLimit(
emails // array to iterate across
,10 // max simultaneous iterations
,checkEmail // an asynchronous iterator function
,function(err){ // executed on any error or every item successful
console.log('Callback of async.eachLimit');
if(err){
console.log('Error: '+err)
} else {
console.log('All emails succeeded');
};
}
);
console.log('Code below the async.eachLimit call will continue executing after starting the asynchronous jobs');
})();
function checkEmail(email, callback){
fetchFromDb(email, function(err, obj){
if(err){ return callback(err) };
sendEmail(email, function(err, obj){
if(err){ return callback(err)};
console.log('Both fetchFromDb and sendEmail have completed successfully for '+email);
callback(null);
});
});
};
function fetchFromDb(email, callback){
process.nextTick(function(){ // placeholder, insert real async function here
callback(null);
});
};
function checkEmail(email, callback){
process.nextTick(function(){ // placeholder, insert real async function here
callback(null);
});
};

Process.nextTick or child_process?

I am trying to allow users to export their contact list in csv format. I am confused on how to run export_connect_csv() function. should i put it in child process or process.nextTick?
function export_connect_csv(user_id, file_location){
mysqlPool.getConnection(function(err, connection){
var csv_row = "Email,First Name,Last Name,Status,Created\n";
function processRow (row) {
var csv_row = row.email+','+row.first_name+','+row.last_name+','+row.status+','+row.created+"\n";
fs.appendFile(file_location, csv_row, function (err) {
if(err){
throw err;
}
});
}
fs.appendFile(file_location, csv_row, function (err) {
if(err){
throw err;
}
var query = connection.query('SELECT * FROM contacts where user_id = "'+user_id+'"');
query
.on('error', function(err) {
//handle error
})
.on('fields', function(fields) {
})
.on('result', function(row) {
processRow(row);
})
.on('end', function() {
//email now
console.log('done');
});
});
});
}
var exportContacts = function(req, res){
var user_id = req.params.user_id || 0;
export_connect_csv(user_id);
res.json({});
};
You don't need to use either, you can just call the function. All of that code will run assynchronously, both getConnection and fs.appendFile. However, you will run into a conflict in the case two users try to export at the same time. You have the following options:
1) You pass a unique file_name every time you call that function
2) You keep things exactly as they are and use fs.appendFileSync to make sure they don't overlap each other but that would block you
3) Or probably the best solution is do what you intended to do with the Process.nextTick, but instead you should use setImmediate and appendFileSync to be able to synchronize writes from several users simultaneously (write only a row at a time to avoid blocking for long periods):
setImmediate(function () {
fs.appendFileSync('filename', JUST_A_SINGLE_ROWW)
});
This is because a recursive process.nextTick can starve the event loop and effecively block you (hence the use of setImmediate) and you need to use fs.appendFileSync because two users might write to the same file simultaneously.
More on setImmediate vs nextTick:
setImmediate vs. nextTick
More info on appendFile: http://nodejs.org/api/fs.html#fs_fs_appendfile_filename_data_options_callback

Resources