Andrew Kohlsmith wrote: > On Monday 05 September 2005 17:04, Jim Van Meggelen wrote: >> I had heard that Digium is not too far off from their >> replacement of the TDM400, either. It's all vapourware, at >> least from an official standpoint, but I'd still advise a >> short wait, because some of these clouds will coalese . . . > > I hadn't heard anything about a successor to the TDM for > low-density POTS > interface... aside from perhaps something to get rid of that > god-forsaken TJ320 PCI interface.
Well, anything that's God-forsaken, is also Jim-forsaken. I might be wrong about a TDM400 successor, but for Digium's sake, I hope not. >> Well, in retrospect it probably was a bit of a low blow, but >> I don't know many people that trust those cards, and I *do* >> have to reboot my system once per week. If those cards are as >> trouble-free as you are suggesting, I'd say the PR folks had >> better get on it! > > Why do you have to reboot your system once a week? What's > causing it? Have > you taken it up with Digium? What rev of TDM card? It's a rev "H" card. I reboot once per week because many months ago I was one of the owners of a product with a problem that affected everyone, but remained officially unacknowledged. A weekly reboot has been a reliable solution, and since I no longer use the product, I don't spend any more time with it. > I'm genuinely curious... I've *never* had those kinds of > issues, and the > rollover bug which plagued the FXO interfaces going "dead" > every 28ish days > has been squashed for a while. There was a six month span during which many people suffered from this problem, and the response was the sound of fanboys blaming the users, and silence from Digium. I very quickly realized that there was no way of knowing when (or even if) the problem would ever be solved, so we put a stop ship on the card. We've found other ways since then, and since the problem no longer affects me, I have wasted very little time with it. Six months was enough for an awful lot of people. I remember hearing that the bug had finally been "fixed", which I found remarkable since, offically, it didn't even exist. Out of curiosity, I had a brief look into this, and if memory serves, for the rev "H" boards, the problem is not actually solved; there is a "work around" if I recall correctly. I've already got one of those, thanks. I simply can't get excited about recommending the TDM400 to people without letting them know about it's storied history. Until a major hardware evolution takes place (including some better method of mounting those daughter cards), label me a sceptic. >> Well, the thing is: knowing what the problem is (or where the >> fault lies) does not make it tolerable. I have never heard >> that multiple TDM400 cards are 100% guaranteed to work. For a >> mission-critical phone system, does that not need to be >> considered? > > Funny story: I have an off-lease Dell system with a Sangoma > S518 ADSL PCI > card and an A101u. I got weird chirping issues on the T1 > ports until I > swapped out the A101u for a T100P. Yep, Sangoma's two cards > were incompatible, but the marginal T100P had no such issues. > Now granted the S518 is just an OEM'd Globespan design but > still... Interesting. What did Sangoma have to say about it? Or Dell, for that matter . . . |-p > I guess the point I'm trying to make is that you test and > retest until you're > confortable saying "yep, this will work." It's the same no > matter what you > use. >From a best practices standpoint I agree, but from a a strategic perspective this had better not be the case. Asterisk, and any hardware that claims to work with it, will need to achieve a far higher level of robustness if it wishes to continue to displace proprietary solutions. Sure, the power of the platform is compelling, and the user must accept some risk when dealing with open source software, but if the basic hardware platforms cannot be made reliable, it's going to be too easy for the folks running the FUD campaign. >> But if you want cheap, "good enough" gear, why not just buy a >> low-end Dell? Save a lot of research and shopping around, >> that would. > > Because Dells are even weirder and (IMO) more of a crapshoot > than the cheap > vendors such as MSI and ECS. There have been many reports of > hardware "sort of" working with Dell gear, especially the > older stuff you'd find on dfsdirect.ca or the dozens of Ya, I know. I was just being a smart-ass. >> I'll grant you that I'm proposing twice the cost for a 10-25% >> improvement in reliability. Still, I would never cheap out on >> a power supply - especially in an analog-based PBX; quality >> RAM does make a difference; and why cheap out on the MoBo? As >> for the Seagate drives, silence is golden. > > I dunno, a cheapass power supply not capable of meeting its > ratings will be > piss-poor anywhere. The power requirements for the TDM card > are grossly > overstated but they're done that way to try and prevent > people from stuffing > a no-name $15 "400W" PSU in the system and wondering why it > spontaneously reboots every 5th call. So long as the RAM > meets its timings > it's good in my > books; that is, if it passes the first 10 minutes of > memtest86's default test > I consider it good. > > Silence? Where are you putting these? I've got RAID arrays > in the server > room louder than anything * will generate. :-) Well, again it's a matter of degrees. I think Seagate has a minimal quality advantage. That's worth a few bucks to me. Noise troubles me because it doesn't come from nowhere. Something *causes* that noise. Does it matter? I don't know. But I like Seagate. >> Having spare parts on the shelf is reassuring, but having a >> system that will go the distance is far more reassuring. > > Oh, totally, totally agreed. > >> Well, when it comes to custom-building PCs, everyone has >> their own ideas about what is worth paying extra for. > > Yup. I wouldn't mind using AMD... it's cheaper, runs cooler > and in theory > should work exactly the same, but I've never gotten good echo > cancellation out of them. I don't know if gcc's just having a > time trying > to write good > code for it or there are nastier things (odd timings, etc., > although that > would be VERY hard to accept) but it's just never worked well > for me. Now are these the 64-bit jobbies; or the Athlons? Jim. -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.344 / Virus Database: 267.10.18/89 - Release Date: 02/09/2005
