Network Connectivity

Tags:
Networking
Help, please! I run a small domain with 4 servers (Win2003)and approximately 60 users. I have 1 machine that is the DC, runs Exchange Server and DNS. It also hosts the databases that run our organization. My problem is that intermittently, machines on the network will temporarily lose the ability to retrieve data from the server. This situation is transitory, does not happen on a regular schedule and is very annoyingl. The recovery time is usuallt 10 - 15 minutes. I have searched all over and cannot find any literature to help solve this problem. Any ideas would be greatly appreciated. Thanks in advance.
1

Answer Wiki

Thanks. We'll let you know when a new response is added.

Is there a trend to which pc’s are having the problem? What is the networking hardware that is connecting the computers (NIC’s and switch gear)?

Discuss This Question: 22  Replies

 
There was an error processing your information. Please try again later.
Thanks. We'll let you know when a new response is added.
Send me notifications when members answer or reply to this question.
  • Ghigbee
    You may also want to consider capturing performance statistics on the server for an extended period of time to see what the server is experiencing during the outage. Maybe there is some job on the machine that is crippling it...just my $.02 but hope it helps.
    0 pointsBadges:
    report
  • Jaysea
    You may have a NIC going bad. I recently had the same problem. Install a second NIC in the server (Static IP)and run it to the switch as well and see if the problem remains. If not replace the first NIC.
    0 pointsBadges:
    report
  • Csmric
    All PCs are affected. The entire domain. We use a variety of NICs, but all are 10/100 and have been checked. We use 3Com 4400 10/100 switches.
    0 pointsBadges:
    report
  • Jaysea
    How many NIC's do you have in the sever ? If you only have one install a second.
    0 pointsBadges:
    report
  • Csmric
    The server has 2 NICs and i have tried them both. Performance logs and Event Viewer show nothing unusual happening during or leading up to the incidents.
    0 pointsBadges:
    report
  • Maeven
    I had this problem and it was a bad UPS on a Switch. I would check those first. Plugged my Switch into the wall and the problem vanished. Got a new UPS and saved myself alot of frustration. Good Luck and let us know what you find. Wendy
    0 pointsBadges:
    report
  • Andynoff
    Use Performance Monitor to check CPU Utilisation/Available RAM/Disk Activity on the DC. Check for network availability by running a PING script from the DC every minute. IOW, maybe either the DC is too busy to respond or it's trying, but can't.
    0 pointsBadges:
    report
  • TJMann
    We also use 3COM switches. We noticed problems if the server NIC and port settings on the switch are not set correctly. Set both sides to either 10/100 FD or to auto. I would also ensure the workstations are also set the same way to improve performance. Hope this helps.
    15 pointsBadges:
    report
  • Bobkberg
    Well, Andynoff beat me to the first punch - by suggesting the regular ping as a means of record keeping. You guys are on your toes. One thing that just struck me - your original question dealt with the ability or inability to retrieve data. Would you be so kind as to clarify - do you mean just retrieving database data? or do you mean complete network connectivity between the workstations and the server? I'd already written more stuff on network connectivity, but thought that we should get down to defining the precise problem before trying to throw any more solutions at it. Bob
    1,070 pointsBadges:
    report
  • Microsolve
    Hi all, Regarding data access, if total network failure is the actual problem (ref Bobkberg's comment) I have a bit of an off the wall thought on a possible cause. If no obvious trend occurs on the downtime, try keeping a record of exact times and dates of the outages. Then try examining all other electrical kit in or around your network cables or equipment. Just occasionally I have read on "blooper" pages of things such as vacuum cleaners or electrically emptying toilets (!) causing so much interference because of cable placement (and possibly quality, but that is another issue) that they take out the network for short periods. A bit of a pig to troubleshoot, but if it helps, good luck!
    0 pointsBadges:
    report
  • Melenie
    Hi, I have read some of the comments regarding your situation and agree with most that it probably is the NIC, however, not on the PC's, but the server. Unless you are using clustering or there is another reason why you need dual nics on the DC, I would disable one of the NIC's and see if the problem persists. For some reason if there is no set purpose specifically for each NIC, like one using DHCP the other for Exchange traffic only. They seem to "compete" with each other and sometimes clients are dropped from seeing the server totally. It is worth a try if it will not hurt your other processes. :)
    0 pointsBadges:
    report
  • GVamos
    The power issues in buildings is more of a factor than people realize. At one of the buildings I worked at the power was transformed 3 times and one of the 3 legs of power was always low. That and the amount of computer equipment in our building gave us very questionable power. Not only was it low because of this 1 of the legs being low but there was "feedback" in the power from so many computers on a small network. We noticed some improvements we upgraded the UPSs to include power conditioning. So power being the problem isn't a myth. Another thing I noticed is that even though we left hubs behind collisions aren't a thing of the past. We were using Cisco switches in conjunction with some Dlink workgroup switches and 3com network cards. The Switches would occasionally get out of sync and then that whole area would experience network issues due to collisions. The 3com nics also experienced issues once in awhile unless the media was forced. Connectivity issues can be even more confusing if dealing with gigabit cards. I ran into an issue that when a machine was added to the domain the network card tried to autodetect speed and momentarily changed speeds during the process corrupting it. So everything looked fine at first but than later issues with connectivity could be traced back to that. I don't know if any of that helped. I more or less just wanted to point out that if the issue goes on long enough it may be time to look at the specifics in detail, not just the general view.
    0 pointsBadges:
    report
  • DrillO
    I can't help but agree with most of the above. A NIC problem can be a real pain. I would like to comment further on the power issues though. I have experienced, over the years, a vast array of power related problems, and I run a really small shop. I have three servers and about 24 users. Power has still been an issue. I was fortuante to be able to free up a sub service in my Electrical Room and dedicated that entirely to my Server Room. I have installed everything with dedicated ground recepticles (the orange outlets)which provides seperate grounding for the circuit and the equipement. I have also installed line conditioners in front of the UPS's so that the sine wave remains constant. I can't tell you how many times I've run into power related problems and torn my hair out looking for other causes. These problems can take up a lot of your time but must be resolved. Good Luck.....let us know how you make out.....it could be something we haven't thought of yet. Paul
    15 pointsBadges:
    report
  • Devilsown
    Another area to investigate in your shop is the power service coming into the facility - when my T1 was installed, the phone company had begun new site testing. They used an inductive ohmmeter to measure resistance on the service/panel grounds. Optimal is around .5 Ohms - mine varied from 25 Ohms to over 150! The old buildings in my area (an old resort town) are infamous for power outages and voltage drops - I used to hear beeping UPS's echoing through the halls daily. Our skilled electrical contractors came out and regrounded everything, referencing the ground @ the power pole. The resistance on the ground dropped to .25 Ohms. That, and the addition of power conditioning in my comm room, has virtually eliminated our problems (constant brown-outs, T1 interface "freak-outs", fried telecomm equipment after every bad storm, etc.). Good luck! m/(-_-)m/
    0 pointsBadges:
    report
  • Guardian
    If you have both then it means you are trying to or already have a VPN?Secondly what error appear cause if there are no error both on the PC's or Servers (which most servers highlight)then try checking if you switches are constanly powered on (but with XP And 2000 OS's they highlight that media not connected)? What are the two NIC's for?And what IP addresses are assigned to them? Are you using DHCP,WINS,DNS and what are their IP addresses,cause sometimes automatically assigned addresses can be of a different range.Check tempretures for all areas that have PC's including the Server room Regards Newton
    900 pointsBadges:
    report
  • Larrythethird
    You could use an open source monitoring tool, like Nagios, to get trends and outages. It could tell you who is failing and when.
    0 pointsBadges:
    report
  • Csmric
    Hi, All. Thanks for all the input. Some of the suggestions are ones I have already tried. the power issues we addressed a few months ago with line conditioners, multiple UPS and isolated circuits. The NICs have been checked and are good. BTW, the machine has 2 NICs because it used to serve as a dual-homed server that provided our network with connectivity to our ISP using ISA Server. That role was transferred to another box and this server was given a fresh install of Windows 2003 Server. Some of the other suggestions are ones that require interupting service and I will try this weekend when the network can go down (basically an 8 - 5 shop except for me). Just wanted to update all you wonderful people who have taken the time to give me guidance that is very needed. Thanks and have a great weekend. Ric
    0 pointsBadges:
    report
  • Vic8man
    Here is an odd one - with the two NICs on the server we had them connected to a single Cisco switch. The clustering of the NICs had their MAC addresses in there lookup tables. For some reason the NIC doing all the work provided the MAC of the other NIC and it comfused the switch into having two NICs with the same MAC address. Data no longer flowed. Reset the switchport (disable / enable) and all was fine. Other Servers with clustered NICs had each NIC on seperate switches and they never had that problem. Only saw it once, but may be check the switch ports for anomolies as well.
    0 pointsBadges:
    report
  • MrOperator
    You need to upgrade your network to be faster. If you have 10/100Mbps for the network, you need to upgrade your network a piece at a time to 1 Gbps. This is a suggestion or check intermittent problem in your patch cables. You could have too many people getting on at once. You can get people to sign on about 5 minutes from when the last users signed on. What is happening is that too many people are signing on at once and creating a bottleneck. You have probably squeezed all the bandwidth to the fullest and fastest you can.
    0 pointsBadges:
    report
  • BigBadDen
    You can have both interfaces connected to the Cisco switch for the purposes of load balancing. You just need to set up an etherchannel first which will enable the switch to use one mac for the aggregated switch ports.
    0 pointsBadges:
    report
  • Csmric
    Thanks to ALL for thier timely and useful input. This last weekend I did several things based on suggestions I received. 1. I changed the NIC and the port on the switch to 100FD rather than AUTO. Since the port is 10/100 and the NIC is 10/100/1000 this might help. 2. I disabled the 2nd NIC. Although it was not in use, I did this anyway to ensure no problems. 3. Rechecked all electrical. Line conditioners are in place and working at the panel. This specific box has dual power supplies each plugged into UPSs that are on separate circuits. The switch is also on a line-conditioned, UPS-protected circuit. So far there has been no problem. I am keeping my fingers crossed and thanking my lucky stars there are lots of people like all of you willing to help strangers. Thanks Again, Ric
    0 pointsBadges:
    report
  • LightSQR
    I have seen this issue on our network, and by changing the duplex and speed settings manually on the NIC it was resolved.
    0 pointsBadges:
    report

Forgot Password

No problem! Submit your e-mail address below. We'll send you an e-mail containing your password.

Your password has been sent to:

To follow this tag...

There was an error processing your information. Please try again later.

Thanks! We'll email you when relevant content is added and updated.

Following

Share this item with your network: